53.3.»... 547...... 3%.? «9... 5r 93%....36» ... any.” a. H.1.th.fl.£¢ . . ‘rhnwt‘fi . , . . . . . Hfixmm. . 8,.“ . . . . .. . v . . . . . . , . tuna... vl Hbl'vu'54 . . .. . . . . . v . . Lilltnu...‘ ,. . . . . . . . i , I?! . . . . . :1.an A. 3.53..) 5... . . 3:. al»...I 7.9m”. 4.....Ir ......w....»4...sf1n—: .Pl )3. , .2346. “paluLhHII’Il. ‘. .fii ’W‘." “Le... 019.....«afl ! a: .11 . .22.: . .- .. ”Wuhan”. 6-H... n.‘..\. (v.5.i...r . . [‘32I‘I. . 4 .....,..........,......¢.H...C.,.. . . gig? . , If , . . L 1 . a viz-Bowmuwwuagufia J, . 7 n a _ ‘. .yy‘“ ‘WQ' ?. 'f - “v'. -/ A~-.-.‘- 333:1 State Uaivcrsity i“ BINDING m . “W ‘J. HUN: 8. gang M" _ .M 8173K BINDER-.1 tar; Ei‘. ' 1...... 1 , ‘ SPFYNGPDRT ABSTRACT A COMPUTER-INTERACTIVE BIPOLAR PULSE CONDUCTANCE SYSTEM By Keith Joseph Caserta A computer-interactive conductance measurement system has been developed which utilizes the bipolar voltage pulse technique for con- ductance determination. The entire measurement system is controlled by.a dedicated minicomputer. Within the system the computer exercises c00trol over the pulse width, the pulse height, the amount of offset current applied, and the tracking amplifier gain. The computer also cOutrols the triggering of the bipolar pulse perturbation and the pulse repetition frequency. It monitors the conductance signal produced by the sum of the cell and offset currents, and stores and analyzes this data. Discrete conductance measurements may be made as often as every thl'Ir'ty microseconds. The dynamic range of measurable conductances extends from 0.22 to 1.8 x 10‘79". Signal-to-noise ratios for this instrumental system vary from 5°40 x.lO2 to 6.70 x 103 over the operating range, for single bipolar pu‘Seeperturbations. Averaging up to 2000 perturbations per point 1“creases these values to l.l8 x 103 to 6.40 x 105 over the operating range. The conductance measurement is unaffected by parallel cell caDacitances at least as high as 1000 pF. Accuracy within the operating 53$ Keith Joseph Caserta range has been found to vary from 0.38 to 0.0037 percent for a series capacitance of l0 uF. Increased accuracy is obtained at larger series capacitance values. The conductance system is capable of measuring temperature simultaneously by means of a separate analog temperature monitor and digital conversion circuit. Software has been written which utilizes this measured temperature for correction of conductance data for tem- perature fluctuations. This is done by curve-fitting a temperature- conductance profile obtained by changing the temperature of the system 'u>be studied over the temperature range of interest, while acquiring conductance and temperature data. The coefficients of the fitted curve are used to calculate temperature-corrected conductance within various data analysis routines. In addition to the correction features, a 'ficord of the temperature-conductance behavior of a particular system any be obtained. This behavior has been investigated for several ehittrolytes in aqueous media, and several in non-aqueous media. The curves thus obtained have indicated that the temperature coefficient 0f conductance will often change sign over a few degrees temperature Change. This preliminary work has indicated that such profiles may be both qualitatively and quantitatively useful. Other programs within the computer-interactive conductance system SOftware set enable the system to acquire and analyze data for conducto- metric titrations, stopped flow kinetic experiments, dissociation conStant determinations, absolute conductance measurements, chromatography m0“'itoring, instrumental performance characterization, and instrumental Se‘f-testing. In addition, a computer-interactive instructional package has been developed for teaching operation of the system. It includes Keith Joseph Caserta text messages, interactive dialog, and graphic displays. This facility provides the novice operator with most of the information required to Operate and understand the system. It also provides a continual reference source for the more experienced user. A number of aqueous and non-aqueous titrations have been investigated with the system. The chemiluminescence reaction of luminol with base in DMSO and l:l DMSO-EtOH has also been studied with the system by monitoring the conductance, temperature, and chemiluminescence changes. The computer-interactive conductance system has been shown to be a versatile and powerful measurement tool which is capable of monitoring those conductance changes which are too small for, too fast for, or beyond the operating range of conventional instruments. A COMPUTER-INTERACTIVE BIPOLAR PULSE CONDUCTANCE SYSTEM By Keith Joseph Caserta A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1974 I dedicate this Dissertation to my sons, Keith and Kevin. May they love to learn, be strong but gentle, and be free to dream and create and be happy. ii up 51‘ 4' l |IOIf. Ibll f u .. .,P; a ‘o‘ '"IOCI d. :.:.. ‘a a". -.'|. ’h n l' D- I! I p : 'U‘l ACKNOWLEDGMENTS I wish to express my sincere gratitude to Professor C. G. Enke forifis help and guidance during the entire project presented in this Dissertation. Professor Enke succeeded not only in providing technical aid when it was needed but, equally importantly, helped me develop an appreciation of the broader perspective of analytical measurement and scientific technique. I wish to thank the fellow members of Professor Enke's research group for the interaction and comradeship which developed between us mnjng my studies at Michigan State. I especially want to thank Brian Hannand Tim Nieman for their friendship, interest, and help during the past three years. I wish to express appreciation to Professor S. R. Crouch for his aid during the project, and the other members of my (hfidance Committee for their comments and interest. Professor R. N. Hammer deserves special thanks for his contributions to my overall graduate education. I want to thank my parents and grandparents who continuously provided the guidance and encouragement which lead to whatever successes I've enjoyed in my studies and research. Special thanks go to my father who has supported me in everything I've done, all my life, and who probably knows me best of all. Most importantly, I want to express my deepest thanks to my beautiful wife, Connie. She has stood with me throughout these last four years of study, during days when my work has kept me at the laboratory and left her alone with the children for long hours at a time. She has been a daily inspiration to me to pursue my work with determination. And, finally, she iii typed the initial draft of this entire Dissertation while taking care of our two sons and our apartment, and preparing our meals and our move to a new city and a new life. iv TABLE OF CONTENTS Page LIST OF TABLES ix LIST OF FIGURES x INTRODUCTION 1 A. User-Oriented Laboratory Instrumentation l B. Expanding the Application of Classical Measurement Techniques Through Laboratory Automation 3 CHAPTER l. General Description of the Computerized Conductance System 6 A. Philosophical Overview 6 B. The Computer-Interactive Conductance System 7 C. The Computerized System Employed in the Computerized Conductance System 8 D. The Computer-Interactive Conductance System Block Diagram 10 E. The Computerized Conductance System Software Library 13 CHAPTER 2. The Analog Measurement of Conductance 20 A. Early use of the AC Bridge Technique 20 B. Later Improvements in the AC Bridge Technique 23 C. The Bipolar Pulse Technique 26 D. The Analog Circuits of the Computerized Conductance System 32 E. The Power Suyplies of the Computerized Conductance System 40 CHAPTER 3. Implementation of a Dedicated Computer for Total Digitization of the Conductance Mea- surement Sequence 43 Chapter CHAPTER 4. CHAPTER 5. Page A. Interfacing the Digital Computer for Chemical Applications 43 8. Digital Conversion and Control of Analog Signals 49 C. Digital Sequencing of the Bipolar Pulse Measurement 55 Programming the Computerized Conductance System for Optimized Measurement 63 A. Creating a Software Library for Labora- tory Instrumentation 63 B. Laboratory Computer Languages 64 C. Programming with Commonly Provided Languages 67 D. The DEC 05/8 Operating System 68 E. Creating an Experimentally Flexible Software Set 73 F. Determination of the Optimum Measure- ment Parameters for the Computerized Conductance Instrument 74 G. The Preliminary Scan Routine 75 H. The Averaging Routine 80 I. The Timed Data Acquisition Routine 82 Performance Characteristics and Self Testing Ability of the Computerized Conductance System 9] A. Performance Characterization Via the System Software 91 8. Determination of System S/N, Precision, and Resolution 9l C. Determination of System Accuracy 98 0. Optimum Pulse Width Selection l05 E. Scale Change Corrections in the Com- puterized Conductance System l08 F. Linearity, Range and Speed Charac- teristics llO vi Chapter CHAPTER 6 . CHAPTER 7. CHAPTER 8. 6. System Diagnostic and Exerciser Facility Temperature Measurement and Compensation in the Computerized Conductance System A. B. The Temperature Monitor The Software set for the Determination of the Thermistor Response Coefficients of Conductance Conductance Data Enhancement Through Temperature Variation Correction Some Comments on the Shape of Con- ductance-Temperature Profiles Application of the Computerized Conductance System to Titration Monitoring and Analysis A. Specific Software and Hardware for Titration Experiments Titration Data Analysis and Display Software Early Studies with the Computerized Conductance System: Precipitation Titration of Ag+ with KCl Some Observations Concerning the End Point Phenomenon Associated with Certain EDTA Titrations Determination of Small Amounts of NaOH in the Presence of Large Quantities of Sodium Phenolate Titration of Phthalic Acid in l:l DMSO- EtOH Application of the Computerized Conductance System to Kinetic Studies: Preliminary In- vestigation of the Luminol Reaction A. B. C. Specific Software for Kinetic Experiments Specific Hardware for Kinetic Experiments Previous Investigation of the Luminol Reaction vii Page 111 119 119 126 139 148 152 152 156 161 163 174 180 184 184 187 191 Chapter CHAPTER 9. CONCLUS ION REFERENCES APPENDIX 0. action E. Reaction CBHELP: A. B. C. Selected Program Listings xpmzm-nmonm) Stopped-flow Study of the Luminol Re- Conclusions from the Preliminary Investigation of the Luminol Use of the Computerized Conductance System by Future Workers Creating the CBHELP Program Using the CBHELP Program CBTSLS CBPSLT CBTCLH CBTALR CCLALF CDTALC CFPTLI CCLMLT CDPMLT FILEII CBHELP The System Instructional Package viii Page 195 208 212 LIST OF TABLES Table Page l. Computerized conductance system instruc- tion set. 48 2. Performance characteristics. 99 3. Accuracy for various conductance-series capacitance combinations 104 ix Figure 10 11 12 13 14 15 16 17 18 LIST OF FIGURES Computer System Block Diagram Computerized Conductance System Block Diagram Computerized Conductance System Software Set Flowchart AC Conductance Bridge Circuit DC Coupled Lock-in Detector Conductance Measurement Circuit Bipolar Pulse Conductance Device Schematic Diagram of the Computerized Conduc- tance System Analog Circuits Photograph of the Analog Circuits Photograph of the Conductance Instrument Power Supply Schematic Diagram of the +24 Volt Relay Power Supply POP/8 Computer Functions Available with the Heath EU-BOlE Interface System Signal Sampler and Converter Schematic Diagram Control Circuits Schematic Diagram Measurement Sequencer Schematic Diagram Photograph of the Digital Circuits Compart- ment Photograph of the Control Circuits for Offset and Gain Photograph of the Control Circuits for Pulse Height and the Analog Temperature Monitor Circuit Simplified Preliminary Scan Flowchart Page 11 14 21 25 27 33 39 41 42 47 51 54 57 6O 61 62 77 Figure Page 19 Preliminary Scan Routine Output 79 20 Averaging Routine Flowchart 81 21 Averaging Routine Output 83 22 Timed Data Acquisition Routine Output 84 23 Simplified Timed Data Acquisition Routine Flowchart 85 24 Simplified Reset Routine l Flowchart 88 25 Simplified Reset Routine 2 Flowchart 90 26 Random System Noise Profile (Plot) 94 27 Reduction of System Noise to Quantization Level (Plot) 95 28 Ensemble Averaging Improvement in S/N and Resolution (Plot) 97 29 CBPSLT Program Flowchart 101 30 Percent Relative Error vs. Log (R) for a Series Capacitance of 5.0 HP (Plot) 107 31 Conductance-Time Profile for Cooling of a Sulfuric Acid-Water System, Raw and Scale Change Corrected Data (Plot) 109 32 CBTCLH Program Flowchart 113 33 Test Probe Schematic Diagram 114 34 Temperature Monitor Schematic Diagram 122 35 Temperature Monitor Response vs. Thermistor Conductance (Plot) 125 36 Thermistor Response - Conductance Profiles for Luminol. PotassiumTertiar Butoxide, and Re- action Products in DMSO ( lot) 128 37 CBTALR Array Arranger Program Flowchart 130 38 Array Arranged Data from Figure 36 (Plot) 133 39 CCLALF Program Flowchart 134 xi ,. r: ‘II a ‘7 'a ‘I '- Figure 40 41 42 43 44 45 46 47 48 49 SO 51 52 53 Curve Fitting Program Flowchart CFPTLI Program Flowchart Temperature - Conductance Profile and Cubic Fitted Curve for a Sulfuric Acid- Hater System (Plot) Conductance - Time Profile for Cooling a Sulfuric Acid - Water System, Raw, Scale Change and Temperature Corrected Data (Plot) Conductance Profile for Addition of Concentrated Sulfuric Acid to Water. Raw and Temperature Corrected Data (Plot) Temperature Change Nhen DMSO and EtOH are Mixed in the Stopped Flow. Tempera— ture Change During a Reaction of Luminol in DMSO and KOH in EtOH (Plot) Stopped-Flow Study of the Luminol System. Raw and Temperature Corrected Conductance vs. Time (Plot) Automatic Burette Interface Schematic Diagram Simplified CCLMLT Program Flowchart Simplified CDPMLT Program Flowchart Conductometric Titration of Ag+ with KC1 (PIOt) Conductometric Titration of Ca++ with EDTA in NH H + Buffer Showing End Point Anomaly (P at Conductometric Titration of 2.3 x 10'?! Ca“ with 2.0 x 10-45 EDTA in NH3/NH4+ Buffer (Plot) Induced Conducticator Effect for Titration of ta++ with EDTA in NH8/ NH: Buffer Before and After Scrubbing the ell (Plot) xii Page 136 138 140 141 141 144 145 147 154 157 160 162 165 169 172 Figure 54 55 56 57 58 59 60 61 622 623 64 5H5 (its (57? (iii Conductometric Titration of NaOH with Phenol in the Presence of Sodium Phenolate (Plot) Conductometric Titration of NaOH with HCl in the Presence of Sodium Phenolate (Plot) Conductometric Titration of Phthalic Acid with KOH in 1:1 DMSO-EtOH (Plot) Photograph of the Early Hacker Stopped- Flow Apparatus Luminescence Data Monitor Conductance Curve of the Luminol- Potassium Tertiary Butoxide Reaction in DMSO (Plot) Temperature Profile Due to Mixing two DMSO Solutions of Luminol and Potassium Tertiary Butoxide in an Unthermostated Cell (Plot) Conductance Change with Increasing Con- centration of KOH in 1:1 DMSO-EtOH (Plot) Equivalent Conductance vs. (Concentration)1/2 for KOH in 1:1 DMSO-EtOH (Plot) Conductance and Chemilumenescence Curves for a Reaction of Luminol in DMSO with KOH in EtOH (Plot) Conductance Change for the Reaction of Luminol in DMSO with KOH in EtOH Corrected for Background Effects (Plot) FILEII Program Flowchart Photograph of CBHELP-Generated Scope Display of the System Block Diagram X-Y Plotting of a CBHELP-Generated Graphic Display Depicting Pulse Width Selection Curves Program Set Selection Options in the CBHELP Program u-I-i-I Page 177 179 181 186 190 196 197 201 202 206 207 218 221 222 226 Figure Page 69 CBHELP-Generated Program Flowchart for Temperature-Conductance Profile Measurement and Analysis 227 xiv INTRODUCTION A. USER-ORIENTED LABORATORY INSTRUMENTATION A significant portion of the research effort in analytical and physical chemistry, in recent times, has been directed toward develop- ment of instrumentation and instrumental techniques for chemical analysis and basic physical measurement. This research has not only resulted in faster, more accurate, and more precise measurements, but also in the ability to perform measurements which were not previously possible. Perhaps the greatest problem associated with the use of this new and sophisticated instrumentation is that most of these measurement systems require equally sophisticated operators to insure that they will be utilized to the fullest extent. This situation has resulted in the emergence of specialists in Nm, far IR, interferometric, ESR, and ESCA spectroscopic techniques, as well as in mass spectrometry, conventional voltametry, coulostatics, and so forth. These groups of Specialists have been needed in order to operate these instruments in the various modes which have been designed into them and to interpret the extraordinary amounts of data, which these techniques produce. The conventional laboratory chemist does not usually possess the time or the inclination to familiarize himself with the details of operation and data interpretation for more than a few of these methods. Thus. the “"99" Of the experimentation, which he is able to realize through the use of these technique$, is very much dependent upon his ability to cmmcate his desires and goals to the specialists. Unfortunately. t be necessary "meeting of minds“ often fails to occur, resulting in a 1 poorly designed experiment and only minimal utilization of the power of a particular measurement technique. Problems associated with optimum utilization of an analytical method are not confined to those techniques which have a significant group of specialists dedicated to them. In addition to these, there are numerous other techniques which are commonly used by the laboratory chemist to solve analytical problems. These techniques are rarely utilized to their maximum capabilities because there seldom is an individual present who possesses the necessary expertise to take full advantage of the method. What appears to be needed in the case of the specialized instruments and the less complex but more commonly used techniques is a system through which any operator, with a reasonable technical background, can realize the best measurement possible with a particular analytical technique, for each experiment which he designs and executes himself. Fortunately, the introduction of monitoring and control computers into laboratory instrumentation in the past eight years does, at least, hold the promise of not only greatly enhancing the sophistication of instrumental techniques, but also adapting easily to operators who are relatively unfamiliar with the intricacies of these methods. Nevertheless, this goal can be accomplished only if the computer inter- action is properly implemented, not only from a hardware standpoint, but also from a philosophical one. Since computer-interactive instru- mentation has the ability to perform both unique and delicate experiments as well as to produce voluminous amounts of data, there exists a very real danger of building such instrumentation in a manner in which the overall instrumental system (instrument, computer. peripherals, etc.) '1 '5”? 53 -n ‘ ""Ii wzwe . . 'O'V gt. rn‘ Gr " HI. I. a; i‘hali "C". 3. I .1 FC- .w \.- ' .1... a. .F I ‘0 .. , .3 .1." .1cl‘. I.. . I‘ “k" ‘d' :. I“ has a greatly increased "apparent" complexity (as it appears to the operator) as opposed to more conventional systems. From a Chemist's viewpoint, then, the goal in designing computer- interactive instrumentation should be to permit maximum utilization of a technique. This should be done regardless of the complexity required for any component of the system. However, an equally important consideration is that which requires the measurement system-operator interaction to be such that the experiment performed with the system will app§a[_to be only as complex as the chemistry involved. Further- more, the final form of the data produced by such a system should make interpretation as simple as possible. If any analytical technique is incorporated into such a system, the result will be a user-oriented laboratory measurement device which is not dependent upon operation by a specialist. If properly designed, the device will be sufficiently flexible to move from one analysis problem to another totally different one with little modification beyond that of changing the operating program. The instrument can then be "time shared" among many different users, much as the computer itself can be utilized in the time sharing mode. 8. EXPANDING THE APPLICATION OF CLASSICAL MEASUREMENT TECHNIQUES THROUGH LABORATORY AUTOMATION Recently, there has been a renewed interest in those measurement ‘techniques which monitor the bulk properties of solutions. These 'techniques do not differentiate between chemical species but are responsive to some overall physical property. Conductance measurement, with which this thesis is concerned. is, of course. one of these classical bulk property techniques. It has traditionally been utilized for titration monitoring, where a specific chemical reaction between analyte and titrant is chosen such that a chemically specific analysis method need not be used. Chromatographic monitoring is another area in which conductance instruments are being used as detectors. Finally, as will be demonstrated later in this manuscript, reaction mechanisms can be studied with conductance techniques alone, or by coupling these measurements with other, more chemically specific, methods. Conductance methods, although certainly in widespread use, have been generally ignored by analysts. This is due both to the lack of understanding of interferences present in real systems when these measurements are made and to the application problems associated with the use of classical conductance instruments. The problems include the necessity of platinizing electrodes, delicately balancing bridge and grounding circuits, gross dependence of the property measured on temperature, and the calculations necessary for correction of conduc- tance data for interferents, for analysis of chemical systems, and for plotting meaningful data. The measurement of conductance, although often utilized in laboratories concerned with widely diverse studies, has usually not been performed in an optimum manner, and has seldom been exploited for its maximum capabilities. This is the direct result of the problems mentioned above, as well as the lack of measure- Inent speed from which the technique has traditionally suffered. Thus, conductance techniques fit easily into the category of research techniques which lack a widespread group of routine applications specialists to oversee the modernization of the technique. It is the purpose of this thesis to demonstrate how the application '1 PA . of a dedicated computer to the monitoring and control of the entire conductance measurement process results in a dynamic laboratory measure- ment system which overcomes most of the problems associated with the conventional application of the conductance technique. In addition, the use of the computer-interactive system expands the application possibilities of conductance measurement into areas where it has not previously been used at all, or only used conservatively. Finally, the system described here enables an unsophisticated operator to perform experiments which make the maximum use of conductance measurement and produce the maximum useful data automatically, without regard to the operator's degree of understanding of the intricacies of the measure- ment being performed. CHAPTER 1 GENERAL DESCRIPTION OF THE COMPUTERIZED CONDUCTANCE SYSTEM A. PHILOSOPHICAL OVERVIEW The thoughts presented in the Introduction to this manuscript could certainly serve as a basis for design of any computerized measure- ment system. The fact that they were, as will be seen, important considerations in the design of the computerized conductance system does demonstrate that they are definite principles which can be implemented in the design of user-oriented laboratory measuring apparatus. However, these principles are, in themselves, of little value to the researcher unless he is directed toward the chemical problems which this approach to instrumentation is strategically designed to solve. To one who would be a chemist, the development of instrumentation performing no unique chemical measurement would appear to be a foolish pursuit. The author feels certain that this is the case. The author wishes to stress, at the beginning of the discussion of the actual research presented in this thesis, that the work to be presented here did ggt_involve the building of a unified conductance instrument. The group of circuits which perform conductance measure- ment per se do not constitute an independent instrument. Except for an on-off switch they possess no dials or dial settings and no switches or indicators which would enable them to be operated in the way that any other instrument might be. They constitute several functional blocks out of many separate blocks which, when properly combined in implementation, are the viable chemical measurement tool referred to here as the computer-interactive conductance system, Other functional blocks, such as the temperature monitor, real-time clock, and display devices, are of nearly equal importance when the overall operation of the system in a chemical measurement application is considered. The entire system is welded into a functional unit through a series of interfaces and a collection of flexible software, all of which are rendered interactive through the power of the controlling computer. The author can assume considerable responsibility for the design, construction, and ultimate operation of many of these functional blocks but not, by any means, all of them. However, the unique manner in which these blocks are combined to create a laboratory measurement system which is user-oriented, and the chemical measurements which have been made with this system to the present time, are the direct subject of this thesis and the author's contribution to chemical re- search. 8. THE COMPUTER-INTERACTIVE CONDUCTANCE SYSTEM The computerized conductance system itself is constituted from three interacting elements, the computer and its peripherals, the conductance instrument (referring, now, to the group of circuits which, when placed in the system, perform conductance measurement), and the software. The computer intimately takes part in the operation of the conductance instrument through the logical flow of the programs. The experiment, monitored most directly by the conductance instrument itself, is under the supervision of the measurement system at all times. This four-way involvement, software-computer-instrument-experiment, results “in experiments which are modified during their execution by the measurement system in such a way that the best measurement attainable for that particular chemical process is obtained. Furthermore, the operator, having set up the experiment, need have no further knowledge of the intricacies of the bipolar pulse measurement technique because the system gge§_possess that "knowledge" through the logical sequencing and decision processes performed by both the hardware and the software throughout the experiment and subsequent data analysis. Finally, the computerized conductance system itself, has become capable of supplying the operator with much desired information concerning the theory, operation, adjustment, and trouble-shooting of, as well as program selection for the computerized conductance system (Chapter 9). C. THE COMPUTER SYSTEM EMPLOYED IN THE COMPUTERIZED CONDUCTANCE SYSTEM The computer system which was available for use with the com- puterized conductance system is shown in Figure 1. It has proven to be an extremely powerful system forlaboratory use. This arises from both the standard peripherals it includes and the specialized peripherals interfaced in this laboratory. The computer is a Digital Equipment Corporation (DEC) PDP-8/I minicomputer with 12K of memory. 8K of this memory is standard core memory and 4K is a solid state memory block manufactured by Calcomp Galaxies Inc. The standard computer peripherals include an extended arithmetic element (EAE) for fast, hardware multiplication and division, «dual magnetic tape units (DECTAPE), a high speed paper tape reader and annch (all made by Digital Equipment Corporation) and an ASR 35 .Eecmmro xqum emumxm cmuaneou ._ mg=m_m mmhm4mh - ezmzmbm uHemzzeHm< amazuexm - momo=m= Lo x NP 2%“: w 102:; mmo<4¢m~o mmFFOAA 7x «unmoumx hmzoo , ea mflaém 4oa mucmuunucou maps; Lopoarm < Tm limo .o ucsmvm 28 that RxCx>>T>>RoCP where R0 is the output impedance of the pulse generator. If this is true, Cx will develop only small polarization during the pulse and will charge approximately linearly with time. Thus at the end of the first pulse, CP is fully charged to E] and is drawing no current, and Cx is charged to some potential less than E]. When the polarity reverses, CP will again quickly charge to the new potential, E2 (-E]). Cx will begin to discharge approximately the same number of coulombs that it charged during the first pulse. At the exact end of the second pulse, the voltage across Cx is nearly zero and CP is again drawing no current. Thus the instantaneous current measured at this time is only that current due to the voltage drop across Rx. The effects of Cx and CP are not just reduced but are virtually eliminated when pulses of the appropriate duration are chosen. The measurement itself is there- fore subject only to the limitations of the particular instrument and not to the cell design, chemical application, or solvent system employed. Johnson and Enke showed that the theoretical relative error, 0, for this measurement technique is given by: Q = b(a - 1) - b2[a(d + 2) - 11/2 + o3[a(a2 + so + 3) -l]/6-... where a = -E]T]/E2T2 b = T2/RXCx and d = T1/T2 If the pulses are truely symmetrical (E]=-E2 and T1=T2), Q~-b2. Since pulses as short as 10 useconds can be easily utilized with state- of-the-art electronic circuitry, even for Rx=100 ohms and Cx=10 pF, Qle"4 or 0.01% relative error. There is no theoretical dependence on CP as long as CPR0 (Ro is the output impedance of the pulse generator) <5T2. \ P‘ 1:" '11.: M: I _ I ‘ r1 '4 {A ‘3 k .’ 9 ~—: ‘0. a.) v". F‘f 29 For their prototype instrument, Johnson and Enke reported 0.01% linearity, no dependence on CP (<0.0004%), the predictable dependence on Cx which was always less than that obtained using the conventional bridge, and a dynamic range also superior to that of the bridge. For a later instrument, with analog interactions similar to the instrument which will be reported here, Johnson (18) claimed accuracies of up to 0.01% and sensitivities of up to 1 part per million over a dynamic range extending from 0.10.] to 10'707]. Recently, Daum and Nelson (19) announced an interesting variation on the Johnson-Enke bipolar pulse technique. They claimed to have over- come the problem of series capacitance effects, to which the Johnson- Enke method is still susceptible at low resistances, by controlling the current in the cell through the application of bipolar current pulses. The resistance of the cell can then be determined in either of two ways. The cell voltage at the end of the second pulse (equal to the product of Rx and the current supplied) can be sampled, or the cell voltage during both pulses can be rectified and integrated. They showed that for the second method, the integral was proportional to RX° They chose this method of analysis for its expected improvement in signal-to- noise ratio. They claimed excellent results in the area of conductance greater than 10"20'1 , where the bipolar voltage pulse technique encounters its most serious series capacitance effects. They felt they could obtain accuracies of 10% for conductances as high as 100']. The potential of the cell during the current pulse is shown in the following waveform: 3O 1 - ' ' IT C IR " “Kiri/ex At the beginning of the current pulses, the potential across the cell will rise very rapidly, but not instantly, due to the current required to charge CP to the voltage IRX. Later in the pulse, the voltage in— creases due to the charging of the series capacitance Cx. The charging of CP is a function of the RXCP time constant which is, of course, smaller for small Rx. If the polarity of the current pulse should be reversed before CP is significantly charged, the IRx drop and the amount of charging of Cx will diminish and the voltage at the end of the second pulse will be somewhat less than it should be. The current pulse must, therefore, last long enough to cause the potential change in the cell to be predominately a function of IRx and the charging of Cx(IT/CX). Longer current pulses are required as Rx increases due to the slower charging of CP through larger RX' As mentioned previously, the bipolar voltage pulse technique requires that as the series resistance and capacitance get smaller, the pulse width must decrease to prevent significant charging of Cx from affecting the measurement. This results from the predominant term, b, in the error equation. A decrease in Rx causes 6 to increase. This increase must be offset by a decrease in pulse width, T, to maintain an accurate 31 measurement. There is virtually no effect of the parallel cell capacitance on the measurement. Thus, it can be seen that the bipolar current technique essentially exchanges the Cx dependence of the bipolar voltage technique for a CP dependence. Better results at low Rx are obtained at the expense of measurement speed at high Rx. Both techniques have distinct advantages in particular conductance regions. However, for solutions normally encountered in conductometric studies, the cell resistance rarely is below 1000. One possible exception to this might be highly buffered solutions required in biological studies. Nevertheless, the bipolar voltage pulse technique appears to provide the most useful operating region as well as the theoretical ability to measure more rapidly through- out this region as compared to the bipolar current technique. The difficulties in the application of conductance measurement to real chemical problems, which the non-expert finds discouraging, are largely eliminated through the use of the bipolar voltage pulse tech- nique. The cell design becomes relatively unimportant due to the expanded dynamic range of operation. Since Cx is no longer an interferrent, platinization of the electrodes is not necessary. By clever use of a four cell lead system, Johnson and Enke reduced the effect of contact resistance, making cell connections less critical. However, selection of the various measurement parameters incorporated in the new instrument (18) was certainly time consuming and required significant expertise. Training was required in both the use of the instrument and its applica- tion to solving particular chemical measurement problems. Conductance detenmination, as a step toward chemical determination, had been brought (floser to the experimenter. Nevertheless, the interaction between the -‘on|€rl£r 31:2 cm N;:l A" r‘» up? 2': hstr ‘r;, in 32 experimenter, the instrument, and the chemical process might still be quite complex. 0. THE ANALOG CIRCUITS OF THE COMPUTERIZED CONDUCTANCE SYSTEM The computerized conductance system has been developed to overcome these application problems and the others previously discussed. Since the instrument was specifically designed for computer control and monitor- ing, it was possible to simplify the analog circuits by use of digital circuit adjustment techniques and a digital data acquisition system to be discussed in Chapter 3. Figure 7 is a schematic diagram of the analog circuits. It consists of the precision power supplies, the pulse generator, the offset generator, and the current follower. Because the pulses must be exactly equal in magnitude for the technique to be free of dependence on the series capacitance and because reproducibility iS‘a function of the long term stability of these signals, the sources of the positive and negative voltages were chosen with care. The precision +5 volt signal is produced by a Signetics SS723L regulator. This signal is inverted by operational amplifier A], an ultra low drift amplifier set to give a gain of -1. The precision +5 and -5 volt signals are fed to the two voltage divider circuits of the pulse amplitude control. The computer selects the pulse height to be either :5, t0.5, or 10.05 volts by switching one of the three DPDT reed relays at the outputs of the divider. The pulse generator. operational amplifier A3 connected as a fast voltage follower, applies the chosen positive and negative pulses to the cell. Pulsing occurs during the timing sequence A-B-C, shown on the waveforms in Figure 7. 33 .mpwauewu mopm=< soumxm oucmpusucoo uwnwcouaneoo we» we Eecmora oeueEmgum .n ogamwe 2m ZN Va 5 \ I. m E: + xoe \\\, mmw. ll. _1|L(2)w|nu\\ill ~< .172 EL v.2 CC»: /_ mogmzme 5:8 I? 1313. 352 V. a o . rrhhua rhhhg V- O m , Igld D‘P‘D‘ P4P‘)‘ P4P4)! D4D‘D‘ \ v.2: 5:28 5 _ _ _ _ :, me _ x2 gm v; \1 Eb" EC: E»: E»: mmzojom hzmmmau . > . «.31 H w e... “v o “v M + “—30.0 ‘11 1.. >5 ~z 55:8 .. P .v 8— mu \ .\ F P< n ...\ .1“ .n o - «.9. .w. .e rrhhgu “.0 “.o > > .J\ 1 9 3.2 .523 2, +> . t \ AV AV rt ::E .3. .3. .v a .v m >m_.+ ”.25.sz ~32 4H\ d . \ rrr»; r—OOIDOO NOD 34 The positive pulse is applied when the measurement sequencer causes field effect transistor switch X to change state at time A. The polarity reverses when FET switch X returns to its initial state at time B. The output of amplifier A3 is connected to the cell during the pulsing inter- val (time A-C) by FET switch Y, which is also controlled by a measurement sequencer waveform. TWO cell leads are used here, contact 1A to maintain the cell at the chosen voltage level and contact 2A to supply current for the cell. This was done to reduce the effect of contact resistance by causing that resistance to appear to be part of the input impedance of amplifier A3. The other electrode is similarly connected to the circuit by two cell leads from operational amplifier A4; contact 18 provides the control potential and contact 28 the current path. Amplifier A4 is the very fast current follower and grounding amplifier. It con- trols electrode 8 to be always at virtual ground and sinks all spurious cell currents to virtual ground when pulses are not being applied by its connection to electrode A through FET switch Y. In order to provide additional resolution and improved sensitivity, most of the cell current may be offset by a stable, constant current supplied by operational amplifier A2. The required amount of offset is selected by the computer through the appropriate combinations of feedback resistors. These resistance values are such that 1, 2, 4, and 8 volts (or any combination of these up to 10 volts) are produced at the output of amplifier A2 by opening the corresponding shunt relays. The computer selects a particular decade current scale by switching the signal from amplifier A2 through one of the four resistors provided at its output. These combinations provide offset in integer units from 0 to 10 for current scales of mA, mA x 10", mA x 10’2, and pA. Thus 7 & o I . 35 up to four additional most-significant bits of resolution are added, by the analog circuit, to the 12 bits of the conductance A/D converter. The cell and offset currents are summed at the inverting input of amplifier A4. During the positive pulse, the cell and offset currents are both positive resulting in a large positive current at amplifier A4. Since the current output of the cell is not being monitored at that time, a 5000 resistor is switched into the feedback loop of amplifier A4 by FET switch Z to insure that the inverting input will be at virtual ground by preventing amplifier saturation. During the negative pulse (time B-C), when the cell current is being sampled, the appropriate computer-selected feedback resistance is switched into the circuit by FET switch 2 under control of the measurement sequencer. The available feedback resistances allow a 10 volt output signal to be produced by amplifier A4 for input signals of lmA, lOO uA, 10 pA, and 1 pA. For the highest gain, a "TEE" circuit (20) equivalent to lOMSl is used for greater precision and shorter response time than could be obtained from a lOMQ resistor. The current follower output is divided to give a signal to the sample- and-hold module of the signal sampler and converter, ES, which is 4/5 of the actual output. To the analog circuit, then, the 0-10 volt A/D converter actually looks like a 0-12.5 volt converter. This was done to provide overlap at scale changes which eliminates the need for pre- cise scale adjustments, as will be explained in Chapter 5. The voltage output, ES, is tracked during B-C and held at exactly time C by the signal sampler and converter. The net voltage, ES, produced by the analog circuits can only represent the true signal to the extent which the various components :5..- 01". f» ’- t.) I (:2 .— 36 of the circuit approach their ideal true values. The choice of components is critical to the production of an accurate, sensitive, fast instrument. Some comments about the components used in the analog circuits are necessary for an understanding of how these goals are successfully accomplished. The precision positive voltage is regulated to 0.01% by the $5723L regulator which will drift a maximum of 0.01% per 1000 hours of operation. Amplifier A], which produces the precision negative voltage, does not have to be fast, but it must be constant and steady. The amplifier chosen (Analog Devices 184J) has an initial offset not greater than t50u Volts and will drift a maximum of :3 pV/month and 11.5 uV/°C. Amplifier A2 must be somewhat fast since the offset may be expected to change quickly and settle quickly during an experimental run. It must also be stable as any error produced by drift will directly affect the measured conductance. The amplifier which was chosen (Analog Devices 1488) will slew at 50 V/ usec and settle to within 0.01% of the true value in‘l usecond. It can be trimmed to zero initial offset voltage and has a maximum drift of :50 uV/day and :20 pV/°C. The pulsing amplifier, A3, must be stable in order to provide the true voltage signals to the cell. It must have minimal offset voltage as this would result hiasymmetrical pulses and subject the measurement to series capacitance effects. It must also be fast in order that it be able to switch between positive and negative input signals and settle to its new value quickly. Finally, it must provide sufficient current when cell conductivity is high. At the time the instrument was built, the Analog Devices 1498 had the best of these characteristics with a slew rate of lOOV/ usec, settling to within 0.01% of true value in 1.5 psec. 37 Its offset can be trimmed to zero and its maximum drift is :50 pV/day and :15 uV/C. It can supply up to 20 mA at 110 Volts output. Amplifiers which are still better suited to this use are available now. However, no serious problems have been encountered with the use of this amplifier which would warrant its replacement. Some problems were encountered with the use of this same type of amplifier as the current follower. The current follower is probably the most critical component in the analog circuit. It must not only supply the final composit signal to the signal sampler and converter, but also provide a controlled, stable cell ground. In addition it must track the signal in the time periods of the shortest pulse widths, sometimes being slowed by large feedback resistors. It must also quickly sink all spurious cell currents to ground in the brief interval (as short as 10 nseconds) between discrete pulsing sequences to prevent a build-up of charge in the cell which would result in gross nonlinearity and measurement error. Therefore, a very new, fast amplifier (Analog Devices 50K) slewing at 500 V/usec and settling to 0.1% in 200 nseconds was substituted for the 1498. It can be trimmed to zero initial offset voltage and has a maximum drift of 1500 uV/month and 125 uV/°C. Relays C, D, and E are DPDT reed relays (Electrol RA 30212241) which were fastest available (150 psec on plus 150 usec bounce, 20 usec off) with low contact resistance (0.10). Relays were chosen over FET switches for these applications because the significantly larger contact resistance of FET switches would cause serious measurement errors which could only be eliminated by trimming the circuit at each control point. The “on" resistance of FET switch Z, in series with the selected feedback resistance of amplifier A4, is significant for the 10 KS2 38 feedback resistor. Therefore, provisions are made for trimming the circuit at this point. Relays F-R are SPST reed relays (Electrol RA 30211241), which have specifications identical to relays C-E. All resistors less than 1M0 in the pulse generator, offset generator and current follower are fast response, 0.01% Vishay metal film resistor ($102 or $106). All 1 M0 resistors are General Resistance fast, wire- wound, 0.01% "Nanistors". All FET switches are monolithic N-channel junction FETS with TTL compatible drivers made by Siliconix. In the application where FET switch Z is used, small leakage currents could cause significant errors for measurements of low conductivities. Therefore a special military-grade switch was used with a maximum leak- age current of less than 1 nA. For the purpose of noise reduction through shielding, the instrument itself and its power supplies were built in two separate modules. Within the instrument module, two compartments are arranged, one on top of the other. Most of the digital circuits and the interface, which will be described in Chapter 3, are located in the large compartment. The smaller compartment below it is divided into three sections. The outer sections contain the control circuits and relays which are shielded from the other circuits. The center section, shown in the photograph of Figure 8, contains the analog circuits described above. These cir- cuits are mounted directly above the signal sampler and converter, in the large compartment on the other side. In this way the signal, ES, from the current follower must be sent only a very short distance to the sampler. The circuits in each section receive the signals from other circuits by wires which run through the aluminum shielding walls of each section. 39 >Z>POQ Om<_0mm . ZOOMP __= 5% r“\‘ Photograph of the Analog Circuits. Figure 8. 40 The connection of the instrument to the cell is made by four leads from the analog circuits to gold plated banana jacks on the chassis. A gold-plated "Kelvin Klip" probe assembly (made by ESI) plugs directly into these jacks. This probe clips directly onto the cell leads to complete the circuit. E. THE POWER SUPPLIES OF THE COMPUTERIZED CONDUCTANCE SYSTEM An early production model Heath 181-75 :15 volt power supply card is used to provide 115 volt signals to various parts of all circuits. The +5 volt signal for the digital circuitry is produced by a prototype Heath 181-74 +5 volt power supply card. A Heath 54-206 transformer was used to supply both of these cards. The entire assembly is mounted in the power supply module shown in the photograph of Figure 9. The +24 volt relay power supply is also mounted in this module. A schematic diagram of this circuit appears in Figure 10. It consists of a second Heath 54-206 transformer, a rectifier, and the two transistors which constitute the 24 volt regulator. It can supply up to 1A at 24 volts. It would, however, not have to supply more than 320 mA to the relays at any given time. The power supplies, circuit common, and chassis ground are all connected to the instrument through shielded cable which is grounded at the power supply end only, for noise reduction. In addition, a separate lfigh quality ground line is provided for directly connecting the ground for the precision signals in the measurement to the power supply. Using this technique, noise signals generated on the ground which parallels the power supply lines for the various components has only minimal effect on the precision signals. 41 Figure 9. Photograph of the Conductance Instrument Power Supply. 42 .5235 uBmeozum 333 $38 53% 93> ¢~+ .2 953... l - - eze F>eemoe ILl\ .. a: comp flu ¥m_ 3N .xp x9, _.....Je)(x,\, _ mmomz~ - em. [ BUFFER BAC o-n MEMORY ; BUFFER (MB) , BUFFER - 8MB o-n REGESTER , Z I' L:>{PERATIOHJ I DECODER CORE ”‘0 MEMO v R V ————> IOPl L_L9_:l_T> '09 —L> IOP2 , GENERATOR I —-T-> IOP4 ‘i SKIP LINE :#< 56’ e CLEAR Ac < ETA CPU ‘ INTERRUPT : < .fi OPERATION INITIALIzE . CONTROLLER RUN f’ 'N'T ' > RUN TIMING SIGNAL 9 BT51 TIMING SIGNAL I fl) ms 3 COMPUTER <———— —->WORLD Figure 11. POP-8 Computer Functions Available with the Heath EU-801E Interface System. 48 Table 1. Computerized conductance system instruction set. Device Select IOP (DS) Pulse Function 32 1 Clear conductance circuit flag; clear conductance program interrupt ability 32 2 Gate conductance circuit driver 32 4 Enable conductance circuit program interrupt 33 1 Test conductance circuit flag 33 2 Turn off pulse sequence 33 4 Trigger pulse sequence 34 1 Test temperature circuit flag 34 2 Latch accumulator into relays F - Q 34 4 Latch accumulator into relays C - E, Time base 35 1 Enable temperature circuit program interrupt 35 2 Clear temperature circuit flag; clear temperature pro- gram interrupt ability; gate temperature driver 35 4 Trigger temperature A/D conversion 36 1,2,4 Available to control peripheral devices such as .titrators, flow systems, etc. 2 ‘7 ., 49 B. DIGITAL CONVERSION AND CONTROL OF ANALOG SIGNALS One of the first examples of an attempt to create a general purpose laboratory data acquisition and control system was that described by Lauer and Osteryoung (30), for the PDP-8 computer. Their system con- sisted of an A/D converter, three D/A converters, a real time clock, a relay controller, a solenoid controller, four solid state switches, an interrupt control, and a plotter. They used the system for electro- chemical studies, producing the waveform, triggering the measurement sequence, and tracking, analyzing and outputting the data through the computer. Perone, Jones, and Gutknecht (31) used another system built around an H-P 2115A computer to optimize the measurement parameters in another electrochemical system during actual run time. Keller and Osteryoung (32) were able to perform measurements which would other- wise have been impossible by use of a computerized electrochemical system for pulse polarography. Daum and Nelson (19) demonstrated the only previous use of digital conversion techniques in conductance measurement systems. Since they had chosen to integrate over the entire pulsing period, they were able to use digital counters at the integrator output to provide auto- matic A/D conversion of this signal. Their data could then be displayed directly as a 3 digit BCD nixie tube display or stored as BCD data in MOS-L51 shift registers. This provided them with a digital means of storing data from rapid changes in conductance such as kinetic studies. A D/A converter was also provided in order that the data could be read out of memory and recorded in analog form on a slow time scale. The analog circuits of Chapter 2 were designed for use with computer control and monitoring only. No other means of external Mk ‘I s... eta... 50 control or measurement were provided. In developing an instrument with this type of computer dedication, circuitry which would be found in conventional instruments to provide repeatable pulsing, analog tempera- tures compensation, signal conditioning for output to plotters, integra- tion, etc., could be eliminated since the computer, coupled with several digital circuits, could assume these tasks. Another major advantage of the computerized system is its ability to make a discrete measurement at the completion of each pulsing sequence (each "scan"). Each data point can then be ensemble averaged with any number of subsequent scans for signal improvement, or stored separately to provide a picture of a rapidly changing conductance pattern. In order to provide measurements at this rate, the signal sampler and converter of Figure 12 was designed. The signal, ES, from the divider at the current follower output, is sent, along with the quality analog ground, to the sample-and-hold module of Figure 12. The tracking/holding sequence of this module is controlled by measurement sequencer waveform 2. Thus, at the beginning of the second pulse (time B), the module is placed in the tracking mode and follows the signal produced by the analog circuits. At the exact end of the second pulse (time C), waveform Z returns to zero, causing the module to hold the signal which existed at that time, as required by the bipolar pulse technique. The falling edge of waveform 2 also triggers a 1.5 IIsecond mono- stable. The monostable pulse, initiated at time C, resets the A/D converter on its rising edge and causes conversion to begin on its falling edge at time D. Hhen conversion begins, the A/D status output goes high and remains high until conversion is complete at time E. The A/D converter performs the conversion on the analog signal held by 51 DS 32 44-—J GATE IOP 2 O A/D CONVERTER ADC-U ACCUMULATOR EIN IN yarn 5.8“; k x 5.8“; H “E258 m I a N. . u A. m A. .n . I: _I< J. 1. , _.----.o....m-------:-: .2 58 was thus built into the instrument. Its functions are described below: The time base for the measurement sequencer consists of the 1 MHz oscillator and its 7 decade scaler on a Heath EU-800KC time base card (29). The output of the card is set by the computer through latch 5 of the control circuit (Figure 13). The timing circuit is arranged such that the pulse applied to the cell will be ten times this output. This permits pulse widths in decade steps from 10 useconds to 100 seconds. Only the four shortest are used by the system. The time base output is connected to a 7490 decade counter, wired to give a gated divide-by-five output. Initially, flip flop A is cleared by an INIT pulse at the start of the program.’ Flip flops 8-6 are cleared by their connection to QA' 'The signal at QA is set to "l" which inhibits the clocking of the a 5 counter by the chosen time base. When the computer issues the DS-IOP trigger pulse, flip flop A is set, releasing flip flops 8-6. The signal,QA, becomes "0" which enables the counter to be clocked by the chosen time base. The counter output, in turn, clocks flip flops B-G. Flip flops B, C, and 0 form a synchronous three-bit binary counter. Their outputs are used to inhibit the J and K inputs of flip flops E, F, and G which produce the switching waveforms X, Y, and Z discussed previously. The Q outputs of flip flops B and D are also gated to trigger flip flop A on the fifth count (time C). This signal returns the sequencer to its initial state. Provision is also made for the computer to terminate pulsing by a DS-IDP command at any time. The computer thus retains control over the pulse width and the triggering of the measurement sequence. The only un- certainty in the pulse width comes from the negligible (<40 nsec). PrOpagation delay through the flip flops. There is a delay of up to 59 4.5 nseconds in triggering the sequences due to the DS-IOP signal interval. Also, since the e 5 counter will only clock on a 1 4-0 transition at its input from the time base, an additional triggering delay error which can be as large as one time base cycle is added. This results from the uncertainty in the state of the time base output at the instant of triggering. When the measurement sequencer was originally designed, the outputs of the synchronous counter were simply gated to produce the necessary waveforms. This approach was found to be unsatisfactory. The response of the gates was sufficiently faster than the propagation time of flip flops B, C, and D that the waveforms produced by the gates contained glitches at the transition times of these flip flops. These glitches caused the FET switches of the analog circuit to momentarily change state,greatly disturbing the pulsing and signal tracking. By- pass capacitors failed to remedy the situation. The circuit was re- constructed as described above. No significant glitches are present in it. The computer interface, the signal sampler and converter, the two latches which control pulse width and pulse height, and the measurement sequencer are contained on Heath compatible circuit cards in the large compartment of the instrument module. They are shown in the photograph of Figure 15. These cards all plug into the main digital board which provides the interconnections between them. Thus they can easily be removed for inspection. The other section of the control circuits, shown in the photographs of Figures 16 and 17, are mounted, with the relays they control, in the two end sections of the other compartment of the module. They also lift up for easy inspection. < / o o O . 0 fl 0 O o O I 1"f*~‘~r—-———___________ :ung - _. 3.: - Afi‘ .. vw .—- ‘em——_-.nl _ Figure 15. Photograph of the Digital Circuits Compartment. 61 ,.,,/.._... I‘ I 1332.11.“ \ \ L .2 . z . ,2 .fi 4 a. S P, P damn—WNW 1.033.113? . \l. IvrlaIl-ll‘ _ - _ . _ . .. -. . r C lOll‘JJIl I IIEIIEC‘CVI ' l .IIIHEIE. I'llIl’l'Vl : z \Qll HI: I O gig. s. Photograph of the Control Circuits for Offset and Gain. Figure 16. 62 ,\.\'.\L()(§ Figure 17. Photograph of the Control Circuits for Pulse Height and the Analog Temperature Monitor Circuit. CHAPTER 4 PROGRAMIING THE COMPUTERIZED CONDUCTANCE SYSTEM FOR OPTIMIZED MEASUREMENT A. CREATING A SOFTWARE LIBRARY FOR LABORATORY INSTRUMENTATION From the references discussed in Chapter 3 it is evident that considerable work has been done by chemists interested in simplifying applications of computer interfacing hardware in the chemical laboratory. This work has resulted in increased availability of standard, inexpensive interfaces and digital hardware for laboratory uses. Unfortunately, at the present time only a small amount of work has been done to relieve similar problems in programing the systems which have been interfaced. Consequently the experimenter often spends an amount of time programing which is equal to or exceeds the time spent designing and building the instrument. This situation affected the rate of the implementation of the computerized conductance system as is described below. There are three approaches to solving the problem of creating a flexible software set for a particular instrumental system. The first is for the chemist to hire a programer to perform the task for him. The danger in this approach was mentioned in Chapter 3. It is often more difficult for the chemist to comunicate his exact wishes concern- 109 experimental operation and data analysis to the programer than it would be for him to do the programing himself. Furthermore, should the nececessity of altering the program arise after the programer is gone, the chemist is faced with the double task of decoding the original program and rewriting it to suit his new idea. Nevertheless, this method is definitely attractive, especially for those systems where 63 64 changes are to be infrequent. The second, and most gallant, attempt to solve the programming problem involves the creation of a computer language or subroutine set which is specifically oriented toward laboratory operations. Very few workers have implemented this approach (three references are presented below). Their systems are not yet capable of solving sophisticated research problems, but they do indicate a direction in which computer application programming must proceed if the researcher is to realize the maximum experimentation possible from the time he is willing to devote to measurement system development. The third ap- proach to programming, which will be discussed later, involves the chemist intimately in the software development. B. LABORATORY COMPUTER LANGUAGES The earliest of the attempts to create a laboratory-specific computer language involved a scheme to introduce digital computer applications into undergraduate laboratories. Perone and Eagleston (33) wished to effect this introduction without significant alteration of the sophistication of the experimentsthich might result if the students were also required to spend large amounts of time programming. They therefore developed a series of data acquisition and control sub- routines for use with the BASIC language. They chose BASIC because it is easy to learn, is an algebraically-oriented conversational language, and because it is interactive, interpreting and executing programs line-by-line. It is also available on most computers, providing more widespread use. Their new subroutines performed data acquisition, experimental control, and timing through control of such peripherals 65 as D/A and A/D converters, clocks, and trigger lines. The interactions between experimenter and instrument, through the development of software, were thus greatly simplified. However, their routines did nothing to simplify programming for data analysis or manipulation which is usually a much more complex task. The BASIC language itself does, to some extent, simplify data analysis. However, because it is an interpretive language, available core space in a minicomputer using BASIC is limited. BASIC therefore lacks the power and speed to perform data analysis of the complex systems often encountered in research. The next attempt involved the development of an interpretive laboratory computer language, LABTRAN, by Toren, Carey, Sherry and Davis (34). LABTRAN was designed for use with the ELLA system (35) for clinical analysis. LABTRAN consists of nine statements, each of which causes perfbrmance of a specific task such as pipetting, measuring reaction rate, pausing, and so forth. The commands are decoded by the computer into a series of instructions which performs these tasks. The only instruction which required the computer to make a decision was the instruction to compare analyses to determine whether or not to terminate the run. The comparison criteria were input by the experimenter. The net result was a neat system of perfonming routine analysis. An experiment could be designed by a person with no programming background at all. The experimenter would simply list the tasks (not the decisions!) he would perform if he were doing the experiment himself. The computer acted as an elaborate sequencer. The use of the computer's decision making abilities was neglected as LABTRAN contained no provisions for branching, altering tasks, etc. The research applications of LABTRAN could only include those types of measurements and analyses which 66 must be mechanically duplicated often for bulk data compilation. The most sophisticated laboratory-oriented language which has yet been developed is the MIRACL language presented by Keller, Courtois, and Keller (36). NIRACL (Macro Implimented Real-time Analytical Chemistry Language) is a macro assembler built around a modified PAL-ll assembler for use on a PDP-ll computer with a floating point processor and 8K of memory. It makes use of macros, coded statements which usually are designed for a specific task, in performing laboratory operations. When a macro statement is encountered by the assembler, the statement is translated into a pre-arranged set of machine language instructions and a set of constants corresponding to the argument of the macro. Examples of nine macro statements which MIRACL uses were presented. They included branching macros with arguments which could be data tests, printing macros, assignment macros, etc. An interesting macro was the AT TIME statement which initiated a block of statements to be executed whenever the clock time was equal to the time variable argument of the macro. Complex timing algorithms could thus be created by nesting such statements. The authors indicated three basic defects in the use of MIRACL in research. These were the slowness of their floating point processor (all MIRACL functions were performed in floating point arithmetic), limitation by available core space to 36 variables, and lack of a plotting facility and corresponding macros. These problems were to be remedied in a later version of HIRACL. It became evident throughout the article that the MIRACL system suffered from lack of computer power. Looking ahead toward future developments one might envision a multi-level system, both in software 67 and hardware. Laboratory job descriptions of a series of tasks, including branching and decisions jobs, could constitute a super-compiler labora- tory language. Once the order of such tasks had been determined by the experimenter, a large batch processor would translate the jobs, compile the resulting program, and assemble a machine language program suitable for operation on the particular laboratory mini or micro computer which was to be used for actual operation of the experiment. The assembled program could also be optimized for core space and run time by the large processor. Programming could thus be done in the same modular way in which interfacing is done in the Heath system described in Chapter 3. The amount of time which the experimenter would save as a result would be comparable to the savings realized by using a modular interface as opposed to designing and building an interface each time a different experiment was performed. Such a system does not yet exist but its creation would seem inevitable. C. PROGRAMMING HITH COMMONLY PROVIDED LANGUAGES The third approach to programming is, of course, for the chemist to do it himself. This is the approach which was taken with the com- puterized conductance system. The basic problem is obvious; the ex- perimenter's time is channeled toward software development rather than experiment design. It should, however, be noted that such programming can be greatly simplified by use of a higher level language, such as FORTRAN, wherever possible. Furthermore, after the initial time spent learning software techniques and writing the first programs, it was found that the experimenter became sufficiently "fluent" in the language 68 employed that the time required to create new programs or alter old ones decreased considerably. For example, all operating programs for the computerized conductance system presented in this thesis, with the exception of the data treatment program for determination of S/N dis- cussed in Chapter 5, were completely written or derived from extensively modified programs in only four months. This was, of course, the direct result of over a year's previous exposure to programming the system, which constituted the necessary learning experience. As a consequence one of the most sophisticated and intricate program sets in the com- puterized conductance system was created by extensive modification of existing programs in a single day. This was the program set written for simultaneous acquisition and analysis of conductance, temperature, and luminescence data, discussed in Chapter 8. This indicates that once the chemist masters the intricacies of programming, the subsequent time devoted to programming becomes reasonable compared to the ex- perimental time, when a sufficiently powerful, general-purpose compiler system (such as the DEC 05/8 system described below) is available. D. THE DEC OS/8 OPERATING SYSTEM All programs for the computerized conductance system were written using the Digital Equipment Corporation 05/8 operating system (37). This system is based on a Keyboard Monitor which allows the user to control the flow of programs. A Symbolic Editor (EDIT) is provided for creation or modification of source files. A Peripheral Interchange Program (PIP) enables the user to transfer files between system devices. An absolute assembler (PAL-8) and loader as well as a relocatable 69 assembler (SABR) and loader are included. FORTRAN II is available. The FORTRAN compiler translates FORTRAN source files into SABR, per- mitting mixing of these two languages. This enables the programmer to write the necessary instrumental control and monitoring routines in assembler language, while performing data manipulation and system device I/O in FORTRAN. This combination of FORTRAN and SABR is exclusively used in the computerized conductance system programs presented in this thesis. The 05/8 system does contain several idiosyncracies which were discovered during development of these programs. The more important of these are discussed below to prevent future workers desiring to expand this program set from repeating mistakes or encountering problems which have previously been resolved. Because of the limited core space available in the 12K PDP-8/I it is impossible for both data acquisition and analysis routines to be resident in memory at the same time, except for the few very simple programs. It is, therefbre usually necessary to write the data obtained during an acquisition onto tape at the completion of a run, for later analysis. Initially, it had been decided to write file-structured data blocks on tape by use of the device independent I/O command DOPEN in FORTRAN (38). The acquisition program would then use the CHAIN command (38) to call the first analysis program. For some reason, however, these routines failed to work with the programs which had been written for the computerized conductance system. Consultation with DEC software specialists failed to resolve the problem. Fortunately, the eventual solution provided a better means of bulk data storage than that first sought. 70 It was found that the best method for transfer of data from memory to DECTAPE and vice versa was to use the non-file structured I/O commands WTAPE and RTAPE in FORTRAN (39). With these routines the absolute block number on the DECTAPE where data transfer is to begin is input to the routine as an argument. The computer proceeds im- mediately to that location and transfers the data. No consultation with the DECTAPE directory is necessary, as in the case of DOPEN and IOPEN, resulting in considerably faster data transfer. It is now suspected that the problem with DOPEN and CHAIN arose from the use of certain page zero locations (74 to 103) by the com- puterized conductance system programs. It was discovered (40) that these locations are used by the PS/8 operating system (which preceeded 05/8) for device independent I/O pointers. 05/8 was to have been configured in such a way that these locations would be freed. However, it is suspected that this has not been the case. In any event, there is no desire to return to the use of these device independent I/O routines because of the higher speed at which WTAPE and RTAPE operate. It should be noted, however, that WTAPE and RTAPE can only be used with the TC08 direct memory access (data break) tape controller. These routines will not function with a tape controller, such as the TDBE, which does not transfer data by direct memory access. Transfer of the computerized conductance system to a computer system with such a controller would necessitate use of DOPEN and IOPEN for data transfer. One feature which the FORTRAN II compiler supplied with the 05/8 system lacks and which would prove extremely useful is the ability to use the extended arithmetic element (EAE) in its math routines. This FORTRAN package performs all mathematic operations by means of software. 71 even where an EAE is available. A considerable improvement in speed would result from use of the EAE instead, for fast hardware multiply, divide, and shift functions. Unfortunately, since a SABR listing of the FORTRAN II compiler was not available, it has not been possible to incorporate the EAE into the FORTRAN package. The SABR language allows the programmer to be rather careless in core location assignment and paging, by supplying indirect statements and page pointers, where needed, itself. However, this convenience can inhibit smooth operation in some applications. For example, SABR inserts a CDF 0 (change to data field zero) instruction each time it encounters an instruction which uses the indirect addressing mode. This not only results in loss of core space and time but can be catas- trophic in a program sending infbrmation into other data fields under command of the source program. This problem is circumvented in the computerized conductance system in two ways. One is to define an absolute address pointer on page zero. This pointer is then loaded with the address to or from which data is to be transferred. The actual octal instruction is then used in the program to reference indirectly the page zero location. The assembler is "fooled" without further incident. Another method, employed only in the timed data acquisition routine, is that suggested by DEC concerning optimization of SABR code (41). This is to define a series of indirect statements (e.g. OPDEF TADI 1400) which are equivalent to the POP-8 memory reference instructions but contain an indirect bit. These statements will work, if used sparingly, but they g9 precipitate an (illegal character) error message from the assembler. When such an error occurs in 72 assembly, the assembler will ngt_call the LINKING LOADER to load the program but will return control to the Keyboard Monitor when assembly is complete. Therefore it is necessary to have saved the relocatable file generated by SABR so that the LINKING LOADER may be called by the user to load the program. Once this is done, the program will run properly, ignoring the error statements. The final problem with the use of the OS/8 system, which has not been overcome, is that one cannot make use of the program interrupt facility with it. This appears to result from part of the system sub- routine calls occupying the first few locations on page zero, including location zero. Location zero is the address to which the POP-8 com- puters jump for the interrupt service routine pointers when an interrupt is sensed. According to the DEC literature (42), locations 0 to 6 on page zero in each field are available to the user. However, it has been found in operation that this is not the situation. Thus the computerized conductance system has never been operated in the interrupt mode although its hardware is capable of doing this. Despite these few defects, the 05/8 system has enabled the sophisticated programming of the computerized conductance system, for both data acquisition and analysis, to be completed and implemented with relative ease. It seems most instructive to discuss the particulars of each of these programs in the chapters that present the chemical measure- ments which these programs perform and analyze (Chapters 5-8). However, three routines which appear often in the data acquisition programs are sufficiently important and general to be discussed separately. These programs are presented in the following sections. They are all contained in the example system program CBTSLS in the Appendix. 73 E. CREATING AN EXPERIMENTALLY FLEXIBLE SOFTWARE SET The component circuits of the instrument provide the sequence of events necessary to perform a bipolar pulse measurement of conductance. These blocks are hardware combined only to the extent required by these sequences. This was done with the intention of producing an instrument with the highest degree of internal flexibility, both in combination of functions and in timing these functions. The use of digital measurement and timing techniques places the computer's decision- making abilities within the instrumental framework. This encourages the use of "intelligent" rather than fixed interactions between the various separate circuits. In this way, the experiment itself may be "designed" during run time such that the data received as a result of the software-instrument-chemistry correlation are the optimum data attainable by the technique. If these design philosophies are properly implemented, a system which is user-oriented results. Any experiment, even one designed by the most novice operator, will yield the maximum amount of information because the computer will be programmed to automatically optimize the entire measurement sequence. Thus, if a chemist designs a significant "chemical" experiment, even though he has no knowledge of the intricacies of measurement which the computerized conductance system employes, he is guaranteed to receive data which is of a quality commensurate with the quality of the chemistry involved. The only assumption is that the experimenter have sufficient knowledge of the parameter he wishes to measure to enable him to decide that a conductance measure- ment is suitable. It will be seen in Chapter 9 that the computerized conductance system is even capable of providing this information. 74 F. DETERMINATION OF THE OPTIMUM MEASUREMENT PARAMETERS FOR THE COMPUTERIZED CONDUCTANCE INSTRUMENT For the circuit presented in Figure 6 the cell conductance, GCELL’ in MHOS, is given by: RI RF-ES) RD RINRI EINRv where RT is the total divider resistance (100000), R0 is the divider GCELL = resistance to ground, RF is the offset amplifier feedback resistance, RIN is the offset amplifier input resistance (200000), RI is the offset current producing resistance, RV is the current follower feed- back resistance, EIN is the precision -5 volt power supply level (-5.000‘Volts), and E5 is the sampled voltage (on a scale of 0 to 12.5 volts). It was indicated in Chapter 3 that certain arrangements of the circuit settings (for a given conductance measurement) will produce values of the output signal, ES, equivalent in magnitude. It was desirable to design the software in such a way that each time the computer set the circuit, the minimum instrumental error and noise level and the maximum resolution resulted. Since the tolerance for each component, C1, of the above equation is known, the maximum error in GCELL (corresponding to all component deviations being in the same direction), dGCELL’ is given by: d0 = él GGCELL |dC + QUANTIZING ERROR CELL 1:] ‘Et;“' i 75 The quantizing error is the error due to digitization of the signal as a result of the A/D conversion, as discussed by Kelly and Horlick (43). It is the resolution limit of the instrument for a single measure- ment, equal to or less than iI/Z of the least significant bit (LSB) of conversion. For one discrete A/D conversion of 12 bits with gg_ offset applied, the maximum quantizing error of GCELL is the product of cell current and Rv divided by 8192 (1/2 digital value of LSB). A program was written to solve for the error fraction, dGCELL/GCELL’ for all circuit settings and various sampled voltages, ES. It was found that maximum accuracy was obtained at maximum pulse height, minimum current follower feedback resistance, and maximum offset. The error also decreased as ES increased. Since maximum offset also corresponds to maximum resolution, no trade-off between accuracy and resolution was necessary. In summary, then, the sequence of events for the computer to follow in setting the circuit for optimized measurement is: l) Maximize the pulse height to make ES>12.5 volts if possible. 2) If necessary, increase RV to make ES>12.5 volts. 3) Apply the maximum offset possible to bring ES within the range of 0 to 12.5 volts. G. THE PRELIMINARY SCAN ROUTINE In order to set up the system for data acquisition, it is neces- sary for the computer to determine the conductivity region in which the measurement is to occur and to establish the optimum circuit settings for measurement in that region. To accomplish this, all data acquisition routines in the computerized conductance system utilize 76 the preliminary scan routine for measurement initialization. 'This routine is flowcharted in Figure 18. A complete listing of the routine appears in the CBTSLS program in the Appendix. The preliminary scan sequence begins when a "G" is entered on the teletype. The computer initially selects the 10 u second pulse width. It sets the circuit in its widest possible range. This corresponds to minimum pulse height (RD = 100), minimum current follower gain (Rv = 104). and no applied offset current (RI = 0, RF = 4000). These values are latched into the control circuit. The computer waits for the relays to close, then measures. The measurement is tested to determine if the A/D converter is reading full scale. If it is not, the pulse height is increased (RD increased ten-fold) until the converter reads full scale or the maximum pulse height is reached. If the A/D converter still does not read full scale, the gain of the current follower is increased (Rv increased ten-fold) until it does or until the maximum gain is reached. If, at maximum gain, the A/D converter still does not read full scale, the computer outputs a resolution error, stores the scale settings and takes and outputs the measurement, and proceeds to the next pulse width. The resolution error indicates to the operator that the conductance was too small to permit offset to be applied, thus eliminating the extra bits of resolution which the offset provides. As soon as the A/D converter output becomes equal to full scale, the computer begins to apply offset to bring the signal within the 0 to 12.5 volt range of the converter. The computer sets the proper decade scale of offset (RI = RV/IO), latches in one offset unit on this scale, and measures. The computer continues to apply offset until the A/D converter reads less than full scale. At this time the computer 77’ [IE-Ii ——-rSET MAXIM RANGE: Rn - 102. R! - 10‘. R1 - 0. RE - 4000] . PH. RV. orrss I SHOE] Figure 18. Simplified Preliminary Scan Routine Flowchart. 78 records the settings, measures, outputs the measurement parameters, and goes on to the next pulse width. If, on the initial measurement of the sequence, the A/D converter .had read full scale and the computer could not apply sufficient offset to bring it on scale, the computer outputs a high conductance error, which indicates that the conductance is greater than 0.220-1 and, thus, beyond the operating range of this instrument. The measurement is skipped and the computer proceeds to the next pulse width. Typical computer output from the preliminary scan routine is shown in Figure 19. After the parameters for the first pulse width measure- ment are determined, the computer, having measured the conductance at these circuit settings, outputs the chosen pulse height (PH), pulse width (PW), offset units applied (UNITS), and the current follower feed- back resistance (RV). The computer also outputs the magnitude of the voltage signal from the current follower (ES), the calculated cell resistance (RCELL), and conductance (CCELL). When the sequence has been repeated for the next three pulse widths, the computer measures the response of the temperature sensor, as will be discussed in Chapter 6, and outputs the value. Finally, the computer prints the available measurement options and waits for selection of one of them. The circuit parameters determined by the preliminary scan routine are stored in two ways. The instruction words, which, when output to the instrument, cause the circuit to be set in that particular optimum mode, are stored intact, one set for each pulse width. These are later used to set the circuit in the correct initial state when an Option routine is called. In addition, a single parameter word is 79 TYPE "G” TO START G PH 8 OoSOODEffll VOLTS PH 8 OoIOOOE'DI MSEC UNITS 8 009090E+OI RV 8 OoIDOOE¢06 SAMPLED V 8 90729980Effll RCELL 8 00513876E+04 OHMS CCELL 8 0.1946DOE'03 MHOS PH 8 605000E801 VOLTS PH 8 001000E+00 MSEC UNITS 8 009000E+OI RV 8 O-IDOOEfflb SAMPLED V 8 O074I272E+01 RCELL 8 OoSISZBOEffld OHMS CCELL 8 00194825E'03 MHOS PH 8 005000E+01 VOLTS PH 8 001000E+Bl MSEC UNITS 8 009OODE+DI RV 8 OoIODOEOOO SAMPLED V 8 60721130E*O| RCELL 8 0.514343E804 OHMS CCELL 8 0.194423E‘03 MHOS PH 8 OoSODOE+OI VOLTS PH 8 001000E002 MSEC UNITS 8 O-9OBOEPOI RV 8 001000E+06 SAMPLED V 8 90685425E+Ol RCELL = 00516240E‘04 OHMS CCELL 8 Bol93708E°03 MHOS TEMP RESPONSE 8 00I56494E+Ol OPTIONS! I)AVERAGE 2)RESTART 3)TDA 4)CALL EXIT Figure 19. Preliminary Scan Routine Output. 80 stored which contains the information as to how the circuit was set. This word is decoded by a special subroutine (DCODE in CBTSLS) where the conductance is calculated from the circuit equation. In executing the preliminary scan routine, the computer has optimized the circuit setting according to the sequence described above. If the computer should be unable to perform this optimization, appropriate messages inform the operator. Thus, at the end of this routine, the operator may be assured that the measurement of conductance for that particular system has been initially optimized. H. THE AVERAGING ROUTINE One of the option routines which the operator may always select is the option to average a specified number of scans (discrete measure- ments) for a more precise determination of the measured conductance. This routine is flowcharted in Figure 20. A listing of this routine also appears in CBTSLS in the Appendix. In order to utilize the averaging routine the operator must input the pulse width to be used in the measurement (which is chosen by the criteria presented in Chapter 5), the number of scans, from 1 to 2047, to average, and whether or not double precision data is to be taken. (Double precision data arises from the additional bits of resolution resulting from averaging, as discussed in Chapter 5.) The computer then looks up the instruction words corresponding to the chosen pulse width, sets the circuit, and measures. The measurements are summed as they are taken and stored in two words. When all scans have been taken, the sum is loaded into the EAE and divided. This provides fast. hardware calculation of the average ES. The DCODE subroutine is called 81 I11 ill" filil'l III 1033?. mm IEWQHI’RIWE’HT’RI mmmm‘ I'L‘I'Ill O. I: I ._ I NO YES IKIIII'I (Om-lettfli ”I'll i17- l'l'] fili- 1 3.0 ' I .‘. IoilIlil‘Il IMME- Figure 20. Averaging Routine Flowchart. 82 to calculate the conductance. The average ES, RCELL, and CCELL are output, the temperature response is measured again and output, and the program returns to the option selection. The printout generated by this routine is shown in Figure 21. I. THE TIMED DATA ACQUISITION ROUTINE Most chemical measurements involve the acquisition of data at regular intervals. The timed data acquisition routine (TDA) provides the computerized conductance system with a flexible means of sequencing this experimental interaction, both for conductance monitoring and for control of other peripheral devices used in the experiment. This routine is found in most of the conductance system programs for chemical analysis. It is also listed in the sample program, CBTSLS, in the Appendix. The computer dialog for this routine is shown in Figure 22: the flowchart appears in Figure 23. Initially, the computer requests the total number of data points to be taken. Up to 500 points may be taken with this routine if temperature and double precision data are taken, up to 1000 points if they are not. Next, the time interval between points is requested. This can be any time from 130 ‘useconds to 40000 seconds. The computer then requests the number of temperature measurements to average. If zero is input, the temperature measurement is skipped. The computer then outputs the approximate length of time the experiment will require and requests the address of the first tape block on which to write the data accumulated during the run. The operator inputs the pulse width, the number of conductance scans to average (up to 2047) for each point, and whether or not double precision data are to be 83 OPTIONS! I)AVERAGE 2)RESTART 3’TDA AICALL EXIT AT HHAT PH? (MSEC)8 cl NUMBER OF G SCANS TO AVERAGE! TOO. DOUBLE PRECISION? (I8Y008N)8I TYPE EXPERIMENTAL INFO (CNTRL G TO END)! CONDUCTANCE MEASUREMENT OF A 00000! M SOLUTION OF HCL 8/25/74 AVERAGE OF 100. SCANS! SAMPLED V 8 0060561523E+Ol VOLTS RCELL 8 0052052886E804 OHMS CCELL 8 0.19211230E'03 MHOS TEMP RESPONSE 8 00168457E+Ol Figure 21. Averaging Routine Output. OPTIONS! I’AVERAGE 2)RESTART SITDA 4)CALL EXIT 3 TOTAL NUMBER OF POINTSIIOOO TIME BETHEEN POINTS (SECS)8So NUMBER OF T SCANS TO AVERAGEIZSo THIS HILL REQUIRE ABOUT 008333E*DI MIN FIRST BLOCK TO WRITE! 200. AT NMAT PH? (MSEC)8 cl NUMBER OF G SCANS TO AVERAGE! IOD. DOUBLE PRECISION? (I8Y008N)II TYPE EXPERIMENTAL INFO (CNTRL G TO END)! TITRATION OF 50-0 MLS OF 0.002 M HCL HITH 0.01 M NAOH CELL THERMOSTATED AT 24.0 C 3/25/74 TYPE "G” TO START 6 Figure 22. Timed Data Acquisition Routine Output. 85 PRECISION YES F1 gure 23. Simplified Timed Data Acquisition Routine Flowchart. 86 taken. The operator may finally record information about the particular experiment, for future reference, on that printout. When a “CNTRL G“ is entered on the teletype, the computer looks up the initial settings from the preliminary scan and sets the circuit (refer to the flowchart, Figure 23). It translates the timing informa- tion, sets the clock, and sets up the software pointers utilized during the run. It then waits for a “G" to be typed before beginning the actual acquisition. Once a ”G" is typed, the computer may initiate operation of a peripheral device, such as the triggering of a stopped flow apparatus. It then starts the clock and begins to measure. Averaging is performed as in the averaging routine discussed previously. If double precision data are to be taken, the remainder from the EAE division is saved, to be later re-divided by the analysis routine in floating point format. If double precision data are not requested, the remainder is cleared, increasing storage capacity. In addition, the 12 bit dividend and the parameter word which contains the circuit setting information are stored. If temperature information is to be taken, it is measured, averaged, and stored in the same manner as the conductance data. No provision for double precision is required as will be seen in Chapter 6. The conductance measurement which the computer has just acquired 15 t(fitted to determine if the sampled signal, E5, is within the proper limits . The criteria for these limits were determined by the hysterisis "ECESSary for scale overlap which will be discussed in detail in Chapter 5' If the A/D converter reads less than or equal to 0077, the computer will execute reset routine 1 to insure that the next measurement will be 0“ Scale. If the converter reads greater than or equal to 7700. 87 reset routine 2 is executed in a similar manner. If all points are not yet taken the computer may again initiate operation of a peripheral device, such as the addition of the next increment of titrant in a conductometric titration. The computer then waits for the clock to signal time for data to be taken again and proceeds to measure as before. If more than twice the amount of time allowed between points has elapsed, due to a clock error or the time involved in a reset routine, the computer signals that the measurement timing has been destroyed and halts. If all data points have been acquired, the program transfers the data to DECTAPE and returns control to the Keyboard Monitor. If the conductance has decreased such that the digital value of conversion is equal to or less than 0077, the circuit must be reset according to the optimization rules already discussed. This adjustment is performed in the fastest possible manner by reset routine 1, flow- charted in Figure 24. This routine will cause the offset to be decreased So that the signal to the A/D converter is increased to compensate for decreased cell conductance. If the offset is already 1 unit, the pulse heighi: is increased to enhance the signal. If the pulse height is aIr't-Ialdy at maximum, the current follower gain is increased. Once either the Pulse height or gain have been increased, the offset is set to 10 units which will cause the next measurement to be properly on scale. If the circuit had already been set at maximum gain and l offset unit, the of‘i’set is turned off altogether. Finally, the parameter word is res“ for the new values of the circuit settings, these values are Iatched in, and control returns to the main program. Reset routine 2, I" r‘esetting the circuit to compensate for increased conductance. 88 NO lfli‘lflifii‘ YES SET ES ZERO OFFSET N0 Tinannasmlnnl "0 <‘llliilll.fi’ YES [HEEE‘WMII [11.01333 autumn! libidllfl.‘ {III‘K'I‘LIII - ELIE}. InaaunnIlnlueiulnnagfiunll F1 gure 24. Reset Routine 1. (Simplified) 89 works similarly. It is flowcharted in Figure 25. These routines insure that the optimum measurement will continue to be made during the entire TDA run. A At the shortest pulse width, the TDA routine allows individual conductance scans to be made every 30 11 seconds. Approximately 100 11 seconds are required to calculate the average and store and test the data between points, if no resetting is required. Reset routines require 300 11 seconds to become "effective" due to the relay closing time. Simultaneous temperature measurement requires the additional time Of 6 mseconds per scan . 9O YES . ME CURE SC NO NO INCREASE 0F SET UNIT YES NO MIN We!“ GAIN ? YES Tl 'l ERROR YES ‘_ SAGE mm "0 DECREASE PH F1BUN-e 25. Reset Routine 2. (Simplified) CHAPTER 5 PERFORMANCE CHARACTERISTICS AND SELF TESTING ABILITY OF THE COMPUTERIZED CONDUCTANCE SYSTEM A. PERFORMANCE CHARACTERIZATION VIA THE SYSTEM SOFTWARE An instrumental system, such as the computerized conductance system, in which the computer has complete control over the measure- ment process and circuit setting is, in itself, the most powerful tool in the determination of its own performance abilities. It is possible, through appropriate software, to determine these characteristics with considerable efficiency and precision. Accuracy, resolution, and pre- cision can thus be measured relatively easily over the entire eight orders of magnitude within the Operating range of the computerized conductance system. In addition, the computer may assume the respon- sibility for making optimum measurements with respect to any of these characteristics for any particular region within this range. The trial and error tedium associated with manually setting a conventional inst"ument for determination of its optimum measuring abilities is thEI‘Efbre largely eliminated. Finally, performance characteristics Which may only be discussed in relatively qualitative terms for many 1."S‘tV‘UIIIt-znts may be quantitatively determined with high precision in the Computerized conductance system. DETERMINATION OF SYSTEM S/N, PRECISION, AND RESOLUTION The first performance characteristic which was determined for the cOmpuizerized conductance system was the signal-tO-noise ratio. S/N. 91 92 in various parts of the Operating range. From these measurements, information concerning precision and resolution were derived. To determine the S/N the instrument was connected to a standard resistance with, usually, 1 ppm/°C temperature stability (e.g. Vishay metal film resistors type 5106 or $102). A TDA routine, such as CBTSLS discussed previously, was utilized to make 500 measurements of this standard at discrete intervals. Each measurement could be a Single scan or the average of up to 2047 scans. Standards were used which covered the entire Operating range of the instrument. A special data treatment routine, CCPMLD, was written to analyze the data and 9101: it on the X-Y plotter and the display scope. The program computes and outputs the maximum and minimum conduc- tances measured during the run. It also calculates the standard deviation. 0. as given by: " 1/2 2: (a - sz 0 g iél where n = the number of points, G = the average conductance, and (3.1 Is the measured conductance at point i. Finally, the S/N is calculated from: S/N = é/o The program is also capable Of calculating GMAX’ GMIN’ G. O, and SIN for any continuous group of points in the data set. The operator may then select to plot any or all of these points. Such a plot is 93 shown in Figure 26. This plot represents a worst-case situation for precision in the computerized conductance system. This occurs in the region near the highest measurable conductance, corresponding to the lowest pulse height utilized for bipolar perturbation. The data consist of single scans taken per measurement (i.e. no ensemble averag- ing) . The standard deviation calculated for this situation was 4.6 x 10‘552'1. The S/N ratio was 1180. The obtainable resolution, in this region, with no ensemble averaging is limited by noise and not by the 12 bit (1 part in 4096) resolution Of the A/D converter. During the initial stages of these resolution determinations, the method used for ensemble averaging was the same as that described in Chapter 4 with the exception that the remainder generated from the EAE division was always discarded; only the 12 most significant bits of the dividend were stored for later treatment. At this time, it was suspected that the attainable resolution would always be limited, by "0158. to less than the 12 bits of the A/D converter plus any addi- 1lional most-significant bits supplied by the Offset generator. This quickly proved not to be the case as can be seen in Figure 27. This plot clearly shows the measured signal oscillating between two least- “9'” fi cant bit positions for only 16 averages per point in the con- ductance region around 3 x 10'40'1. The resolution has clearly been limi ted by the 12 bits of the averaging which have been kept for anal.l'Sis. Increasing the number of averages could do nothing to im- prove the signal. However, Malmstadt, Enke, Crouch, and Horlick (44) have Shown that a level of system noise greater than 1/2 of the quanti zation level randomizes the quantization error. making S/N enhancement by ensemble averaging possible. Furthermore, once the Am-MAmpo_v .ePeceea emeez seemzm eeeeem .em ee=m_a Rommv were o 94 3 IS .11. IS _ S. _-eN-opxmm.e mopxmm~w_.F u epeea Z\m P-cm-o_x_mmep e u eeeeee>ea eeeeeeem _-em-opxm~mom.e h a emeee>< Pl ~-o.xmpmom.e sze eez _-c~-o_xmemmm.e szu xez .L_ . P-e~-opxmm e 95 Am-MAmPo_V .Pesea eoPAeNPeeeao o» emeez seemsm to ee_eu=eem .AN eeemee Aommv me?» o . a. e S _ .0 _-em-o_x~Nm o coaxeoemm.. u oseem 2\m .L F-ee-opxmemmm N u ee_eee>eo eeeeeepm _-e -osxokmmm.m u a emeee>< _-me-o_xmem~m.m u szu EL: .I P-ae-o_XMAm~m.m I szu xez w a z I. e e u a o z I. o u P-em-opxmmm.o 96 number of averages decreases the system noise to 1/2 of the quantiza- tion level, increased resolution beyond the number of bits of the A/D converter will result. For example, if 16 12-bit A/D conversions are totaled, a 16 bit word will result. If the noise on the Signal was not greater than 1/2 the value of the least-significant bit when the data were taken, two Of the four bits beyond the initial 12 are significant, providing a l4-bit conversion and increasing the resolu- tion by a factor Of 4. It was therefore decided to store the remainder from the EAE division during the TDA run (essentially "double precision") and re- (iivide it by software during the data analysis routine. In this way the predicted improvement in S/N and in resolution could be realized. 11ae effect of this technique is demonstrated rather dramatically in Figure 28. Figure 28 shows a TDA monitoring of a standard resistance in the region of 1000:2(10'39'1). Double precision data have been taken (i.e. the remainder is saved and re-divided during analysis) for an average Of 2000 scans per point. The two irregularities in the curve were found to be a disturbance caused by the temperature bath heater switch- 1"19 on for 10 seconds, every 35 seconds or so. Although the standard resistor was attached to an aluminum plate with heat sinking grease, a Small amount of initial heating can be observed during the first IO'IS points. After this time, the heating by pulsing and cooling by a" reach a steady state. Neglecting these recognized interferents, for Points 250-400 (2000 scans/point), the standard deviation was found to be 1.58 x 10‘99" yielding a S/N of 624,000. Thus the Option to take double precision data has henceforth been incorporated .L‘ ~c~ I...~u\. .s. 97 AwIMNMNOFV .cowu:_0mmm van Z\m cw w:mEm>ocaEH mcwmecm>< m_nemmcm .wm mczmwm Aommv me_e oo_ ow so as ON 0 _| A _ _ J _ _ d 4! a J mo_xeeoem.e u o_eem2\m _-as-o_xmmmmm.p u cowoaeseo eeeeeeem _-ee-o_xemmem.m u a emeee>< I. _-ee-opxmmmem.m u see eez F-ee-opx.oeem.m u zvo xez ”ooe-0mm maeeea mopxemmpm._ u oeeem Z\m .1 P-em-opxeemme m u ceasesseg eeeeeeem _-e -opxmmmkm.m u u emeee>< .-me-o_xmwmkm.m u sze eez .1 .-ce-o_xepeem.m u szw xez "oom-P meeeoa u. 11111111 .u "1-11-1-1-.. IIITIHIIIIHId. IIII .v 11v I.I+IwIIIIL1LIlIIIIIIIIIIIIIIIIIIIIIIIIIL p ....... a m a - mocmcmecmpc_ gums .aemu ......... I II muonnmm.o QOZQDUF—(ZULLJ m-opxmmm.o 98 into all data acquisition and analysis programs. As shown in Table 2, this region of conductance proved to be the most precise for measurement by the computerized conductance system. For an average of 2000 scans, the noise on the signal totaled only about 1.6 parts per million. In the region of highest conductivity, the S/N was limited by the number of averages which could be performed in a reasonable amount of time. The greater noise is most probably the effect of the relatively higher noise levels on the 0.05 volt pulsing Signal employed here. In the region of lowest conductivity the results obtained during these measurements are somewhat limited by the stability of the standards which were available for use. In «addition, at the very low signal levels being detected for these con- tiuctances, spurious currents through the glass epoxy printed circuit laoard become significant compared to the measured signal. Had this problem been anticipated in the design stages, its effect could have been minimized by placing ground loop foil patterns around the cell contacts on the analog printed circuit board. Since few measurements are made on solutions with a conductivity this low, rebuilding the board to correct this problem has not appeared necessary. Finally, increasingly long pulse widths must be employed, as will be discussed later, at low conductivities, limiting the number 0f averages which can be made in a reasonable amount of time. C. DETERMINATION OF SYSTEM ACCURACY The response of the computerized conductance system was found to be linear for any particular scale setting. Discontinuities, which were encountered at scale changes (which will be discussed later), “cwewcmaxm eo mpeum meek .nceuceum eo xpwpwnepm moF x Po.m NPIOP x mm.m coop m-o_ x o.m ugeucepm eo xpwpwnepm eop x n~.~ PFIoF x mo._ ooom NIOF x N.m mmwoz mop x 08.0 mIop x n~.~ p “IOP x ~.m mcwmecm>< eo “were mo_ x en.m Ppuop x om.m ooom mIOP x m.m mmpoz mo_ x Ne._ ole. x e~.w p oIOF x m.m mcwmeem>< eo weave mo_ x o~.e oFIoF x om.~ ooom m-o_ x m.m mmpoz mo_ x me.p x m~.m P muop x m.m o, uceuceum mo happeneum mo_ x m_.N x mm._ ooom euop x ~.m o. mmwoz mop x em.e x me.m P ¢1op x ~.m nececeum to morpwneum mop x oe.m x em._ ooom «lop x m.m mmpoz mop x oo.m x om.p p «Top x m.m mcwoecm>< eo prep; mop x mm.~ x m~.m ooom m1op x a.m mmroz mop x ou.o x Fe.p p mIOP x e.m mcpmecm>< eo «Pew; cop x om.~ x up.~ ooom NIOF x m.¢ omwoz mo_ x w_.P mIOP x mp.¢ _ «Io— x m.¢ hm owuem mmpoz covuew>mo momecm>< mucepoaucoo vote: -815um 2853 .6 L852 .mowumwcouoecego accessoesma .N m4meetween pulses. 1' 30 msec t>eetween pulses. 105 the relatively poorer accuracies in these areas. All values in Table 3 represent 30 nseconds allowed relaxation time between pulses except at 2 x 10-59-1 and 2 x 10'69'1 where longer re'l axation times were found to increase accuracy. It can be seen that over much of the operating region, accuracies of close to 0.02% or better can be obtained for series capacitances of 10 0F. At lower capacitances, the effects of pulse asymetry are greater as can be seen - Correspondingly, at the higher series capacitances of many real conductance cells ( 20 uF), the accuracy would be even closer to that obtai ned with the resistance standard alone. Finally, no measurable effect on accuracy by parallel cell capaci- tances less than lOOOpF has been Observed. Since real cells normally fall well within this region, any effect of such capacitance has been discounted. D. OPTIMUM PULSE WIDTH SELECTION Two factors must be considered in the selection of a particular WISE width for use in a given conductance region. First, the pulse Width must be Short compared to the time constant for the series RC are"? t formed by the conductance cell as discussed in Chapter 2. Secondly, the pulse width must be long enough to allow the current fOlIOWer to settle to its true output at the end of pulsing. For low CODdUCtances the gain of the current follower must be increased, SIO‘” '19 the response of the amplifier. Thus longer pulse widths must be “38d at lower conductances. CBPSLT provides, through Option 3, a means to determine exactly 106 the optimum pulse width for use in a particular conductance region. I n using this option, the Operator inputs the chosen pulse width and the number of individual conductance scans to average per measurement. A preliminary scan is conducted as with option 4. The operator then inputs the true value of the resistance which is being measured. Using the same measurement options which are available in option 4, the operator first measures the standard resistance, then the RC network, and stores the true resistance and the relative error as an X-Y pair. After each measurement is stored, the value of the next resistance is input and the sequence repeated. When the E option is finally selected, the maximum error is calculated and output. The operator then selects the plotting scale and the data is point plotted as relative error vs. Tog (true resistance). The data may then be stored on tape for the more detailed analysis provided by option 1 or replotted by option 2. A composite plot of data obtained in this way for each of the four ShO”test pulse widths, over the entire operating range of the instru- ment, appears in Figure 30. This particular plot corresponds to a se”93 capacitance of 5 “F. The two factors which contribute to the "m" s a pulse length that is Significant compared to RC and the current fonower settling time, are manifest in the rapid increase in error at Either- end of a particular pulse width region. It can be seen from the 91 ot that, for the best accuracy, the pulse width should be chosen as 1’0110ws: Above 1049'] (<10 K0) the shortest pulse width (0.01 msec) is “58d. between 10'494 and 1.4 x 10'59'I(10 K - 70102) the pulse width = 0.1 msec, between 1.4 x 10'59'I and 1.4 x 10450“ (70 K - 700 m) the Pillse width - 1.0 msec, and between 1.4 x 10'69'1 and the 10"" 107 I.I.I.I FUJI. I owns o.oF n 3a a .ume o.~ u 3a O .Omme _.o u 3a .O; 0.0 to OOOOOeOOOOO meaeem O Let AOVOOO AOOOOO OOOA OON SOP _-II _ ._- .. .. .I.....qume.x..zn—mwe ..wx.. d a 0A H a .ume Po.o n 3a 0 .m> LOLLM m>wuwpmm R .om mczmwu om wanton: mm4<}—H>m & 108 conductance end of the scale (700 K - 80 M0), the pulse width = 10.0 ms ec. A series of such plots were prepared for various other series capacitances. They are all in approximate agreement with Figure 30 concerning which pulse width to select for a particular conductance region. E. SCALE CHANGE CORRECTIONS IN THE COMPUTERIZED CONDUCTANCE SYSTEM The instrument, as initially designed, performed well over its enti re range except in the regions of scale changes. It was found to be impossible to perfectly align the pulse height, current follower gain . and Offset such that there was no overlap or underlap of scales. The problem of underlap was most severe as it resulted in a "dead zone" 1'" Which data points were lost altogether, as the computer caused the instrument to oscillate between the lower Offset, gain, or pulse height SEtti ng full scale conversion and the higher offset, gain, or pulse he19"”: setting zero conversion because of the area where the scales failed to meet. All data taken during the underlap interval were lost. 1" add‘ition, some non-linearity occurred at low voltage level outputs Of the current follower, apparently due to pulse asymetry with insuf- fici ent allowed relaxation time. Both problems were solved by PTOVIdihg over] ap of the scales through the divider at the current follower OUtPut. Hysteresis was provided at both ends of the scale by not perm-i tting an A/D conversion above 77008 or below 00778 without a scale change. Any discontinuity occurring at the scale change point (see raw data, Figure 31) was eliminated by programing the computer - TI 1: h‘.'- 109 A~-O~OPPO AOOON-OOV OOOOOO OeOOH OFOO OO new; uomp Empmxm Lopez 1 uwo< uwcsepsm e eo meepoou com OFFeoca week I mocepuaucou ._m mczmwm Aummv meek omm com om, cop om o q 4 _ .14 O .41 O. IO 4 O Fucmnopxe-.~ eueo umuumccoo u even 3mm UOZDDQF—(ZULIJ x ._ _-em-op mPO 110 to adjust mathematically the data at those points (see scale change corrected data, Figure 31. The computer performs this correction by ca‘l culating the best straight line through the last three points measured on the previous scale setting and from this calculation, Predicting the position Of the next point. The Offset between the Pr‘ed 1cted position and the actual position of the first point at the . new scale setting is used to correct all subsequent data. If fewer I than three points are taken at a particular scale setting, as may I be the case for very rapid conductance changes, scale change correction is not performed by the computer. After implementation of the hysterisis and scale change correc- ti on provisions, it was found that instrumental adjustment was virtually umNecessary. Only infrequent triming of the current follower, to p"‘event non-linear response due to pulse asymetry, power supply adjust- ment, and adjustments for extremely accurate absolute conductance determinations are performed. F. LINEARITY, RANGE, AND SPEED CHARACTERISTICS Many of the operational amplifiers in the analog circuit have Very fast response times and, thus, a tendency to oscillate. A 56 pF ca Dacitor in the feedback loop prevents oscillation of the offset am- p] 1 FT‘ er but would cause the current follower response to become non- 1 i hear if used. The resulting small 10 MHz oscillations of the current f0] Tower are transparent to the measurement and are allowed to occur 8 - 1 “Ce the noise bandwidth of the instrument is upper-limited by the ‘ 111 frequency response of the sample and hold module (500 KHz). The instrument was found to be linear over its entire Operating 8 1 range, which extends from 0.22 to 1.3 x 10' 0‘ . Above 0.220“. the conductivity is so high that maximum offset is insufficient to put the conversion system on scale. Below 1.3 x 10-89.], as previously mentioned, conduction between the copper foil pattern through the glass epoxy printed circuit board becomes significant compared to the measured conduc ta nce . l‘ “I.~. v F i nally, the instrument can be completely reset by the computer and settle to its new settings in 300 nseconds, limited by the relay c1051'19 time. Discrete conductance measurements may be made in 30 nseconds (20 nseconds for pulsing and 10 nseconds for relaxation during "hICh A/ D conversion takes place) at the maximum rate. G. SYSTEM DIAGNOSTIC AND EXERCISER FACILITY EVera during the earliest stages of development of computers and computer systems, software sets were designed Specifically for exercising these 8.sttems and indicating malfunctions within them. It was found that i F such tests WEre conducted routinely, the system gained a high degree Of reliability. This is because many malfunctions can be detected befehe they become sufficiently serious to effect Operation of the entire system. Furthermore, Computer systems can become so complex that they are 1Itev‘nally impossible to troubleshoot in any other way. Thus, when a e°mPUter system is developed, appropriate testing software should a ISO be Written. Such software is standard for the computer part of the PDP‘BI I system described in Chapter 1. It includes central processor ‘ 112 tests, magnetic tape transfer and transport tests, extended memory test, and EAE tests (45). This same mode of testing may be extended to non-standard peripherals such as those developed in a chemical laboratory. The computerized conductance system is, of course, such a peripheral. In the interest of a smoothly running system, then, it was desirable to take advantage of the power inherent in software instrument tests. In this way, again, an operator lacking a feel for instrumentation would be able to implement a software test which would pinpoint a particular hardware malfunction. Often, the problem may be solved simply by tightening a circuit card connection or replacing an integrated circuit chip. The software diagonostics can lead the inexperienced Operator to these causes. If the problem is more complex, the system can inform the user to seek The manner in which the operator is taught the use and The outside help. interpretation of these system tests is described in Chapter 9. Program which performs these tests, CBTCLH, is flowcharted in Figure 32 and l isted in the Appendix. Its function is described below: In order to use CBTCLH to perform instrument tests, the test probe shown in Figure 33 is normally plugged into the instrument cell lead c°""eCtions. This probe provides series resistances and series and para] 1el cell capacitances for use in the system tests. These components are avai lable through switches on the probe itself. A BNC connection for 05C? ‘lloscopic monitoring of the pulses is also provided. For most instrument tests, the moon resistance standard with no series or paranej capacitance is chosen. The program is entered by the operator in the nonnal manner. The operator is asked to choose a P0159 "1‘1““ He n Orma] ‘ly will select one of the shortest which provides the greatest ¥ 1 1 3 «mezHOg gamxu ua4m » hum 2% e o_mmm>z0u omMN F amegmquz oe .u. mwa.e xueqeueo mz<_ w mmezm «mow~z» «Hex—om Xumzu o¢4u p hum .o— x 3m ow mmr two significant figures. Johnson claimed excellent results for l°C liegative temperature deviation correction but positive deviation correc- 1:ions were not suitable beyond 0.25°C. In designing the computerized conductance instrument, it was desirable to keep the analog conductance circuit as simple as possible, "(It encumbering it with those amplifiers and multipliers which were necessary to compensate for temperature variation in Johnson's instrument. If! addition, situations could be envisioned in which a record of the aC‘tual temperature change itself would be useful. Reaction studies in which heating due to a reaction or mixing would affect the system Cheflnistry is one such example. In order to accomplish both of these 121 goals, it was decided to include in the computerized conductance system a separate analog temperature measuring and digital conversion circuit for fast monitoring of temperature simultaneously with, but not affect- ing, the measurement of the conductance signal. The temperature monitor designed for these measurements is shown in the schematic diagram of Figure 34. The temperature monitor consists of an analog bridge circuit and difference detector to follow the conductance of a thermistor, RT’ and an A/D conversion system with a flag. The precision 5 volt signal produced by the $5723L regulator of Figure 6 is used to provide the stable signal for the bridge. The bridge is designed in such a way that a lo KO thermistor will produce a potential difference between points A and B of about 1 volt. The voltage level at B is subtracted from the level at A by operational amplifier A5 which also supplies a gain of 5.1] to this difference. The output of amplifier A5 is connected directly, along with the quality analog ground signal discussed in Chapter 2, to a dual-slope, integrating A/D converter. The MD con- verter is triggered directly by a DS-IOP signal from the computer, "E‘Setting on the rising edge and initiating conversion on the falling Edge. Upon triggering, the STATUS output goes high. When STATUS returns to 10w, it triggers a flag and possibly a program interrupt. The flag C1"‘Cuit is identical to the conductance flag circuit discussed in Ch‘1‘l3‘ter 3. After testing the flag and finding it raised, the computer wi] 1 gate the driver and transfer the data to the accumulator. The analog section of the temperature monitor appears in the photograph of Figure l7. It is mounted directly over the temperature moni tor A/D converter which is located in the digital circuit compart- me "1; . photographed in Figure l5. 122 .Eacmawo uwpmsogum coerce: mcaumcmane HHzH N moH mm mo .«m «camel a uzou some mm ma em AV m ezou l/I/I/u .«zanli < a o\< ¥o_ x0, x_ N_ ¥~.NN V:.$ > m+ zoumuummu 123 Originally, a temperature sensor which consisted of a thermistor in the feedback loop of an operational amplifier was used with the same digital conversion system. It was found, however, that considerably more noise appeared on the signal for this circuit than could be tolerated. The bridge circuit was therefore built. The results were far superior to the previous method. Noise on the signal for the circuit of Figure 34 is lower in magnitude than 3 least significant bits of conversion and can usually be reduced to less than the least significant bit level by averaging 16 single temperature points. Amplifier A5 is the moderately fast Analog Devices l498, identical to the 1498 used as the pulsing amplifier in the analog conductance circuit. The A/D converter is a Teledyne-Philbrick 4109l0, a 0 to + 10 *volt converter which utilizes the dual slope technique to convert in 5 In seconds. For the signals produced by ordinary temperature sensors, a liigh speed of conversion is not necessary but some means of signal averaging is desirable due to the noise on such signals. The resolution 01’ these signals can be enhanced through use of an integrating converter. The gated driver is a prototype of the Heath EU-BOO-JL gated driver (25’) identical to the conductance circuit gated driver. The resistors “58d in the bridge and amplifier are all metal film resistors of 0.1% to‘ erance or better. 'Fwo thermistors have been used for temperature monitoring in the computerized conductance system. One of these is the 44006 Yellow Spri ngs Instrument Company thermistor with a resistance of l0 K52 at 250(:- It is encased in a teflon sheath to prevent electrical contact "1th the solution and to protect the thermistor from harmful solvents. T . “‘3 thermistor has a time constant of 25 seconds in still air and 2.5 124 seconds in a "well stirred" oil bath. (The time constant is the time required for a thermistor to indicate 63% of a new impressed tempera- ture). The resistance change for the 44006 thermistor is 4%/°C. It was used for most studies in slowly mixing solutions where response speed was not critical, but mechanical strength was. For very fast applications, such as temperature monitoring in stopped flow kinetic systems (to be discussed in Chapter 8), the fastest possible thermistor probe was desired. The fastest probe which could be obtained proved to be the 4lA40 Victory Engineering Corporation bead- in-glass-probe thermistor. This thermistor is extremely fast due to its tiny size (0.02" outside diameter). It has a time constant in still air of'l.4 seconds, in a still oil bath 0.11 seconds, and in still water ().055 seconds. Its resistance change is 3.9%/°C. Faster thermistors are available from the same company (up to a time constant of 300 useeconds in still air), but they are not available in a configuration sui table for use as a probe and would be too fragile for use in a liquid f10w stream of high velocity. The response of the temperature monitor, the voltage output of a'"Dl‘ifier A5, vs. the conductance of the thermistor, is shown in Figure 35. This curve includes the range from 17 to 43 °C for the two ther- mistors described above. It can be seen that the response is not linear "1th thermistor conductance, which is proportional to the temperature. This is the result of the design of the monitor analog circuit. The output of this circuit, which is the signal. EIN’ which the A/D converter re - . . Celves, lS given by: 25.55R E = T - l 062 IN ‘l‘2‘1"‘“00+RT ' 125 v .wucmpuzuc ou LoumPEc . m; . e m> wmcoammm Louecoz . mezumcmaem H .mm we 3: u — .N x . RV mu m uavco Lo x 4 W umeLw 3 33 _ U . muop ¢F.m 1 Nm©.p m o h H m z h o 4 Z o > m m m a m H z < o m a m m a m z a m k vmo.o 126 The voltage, EIN’ can also be seen to be inversely proportional to temperature. This non-linearity and inverse proportionality have no effect on the implementation of temperature correction in the computerized conductance system. This is because the variation in conductance, as a function of thermistor response (which is the "monitored" function of temperature), is fitted to an equation suitable for describing its behavior. The coefficients of thermistor response are then used in data analysis routines to provide the necessary temperature variation correc- tion of the measured conductance. The details of this correction technique are given below. B. THE SOFTWARE SET FOR THE DETERMINATION OF THE THERMISTOR RESPONSE COEFFICIENTS OF CONDUCTANCE In order to perform corrections of measured conductance for changes If! temperature in the computerized conductance system, the actual varia- tiibri of the conductance of a system with temperature variation must be measured in the absence of other effects, such as changes in the ionic Chair‘acter or dielectric properties of the solution. If these other Effects should occur the resulting correction parameters will be formed to iliclude them. Implementation of these parameters will eliminate the othEY‘ effects as well as temperature effects. Since changes in ionic Char‘élcter or dielectric pr0perties are most often the effect which the Opera tor desires to follow, such a "correction" for them would destroy the measurement. ‘To construct a temperature-conductance profile, the temperature Of ”tiles chemical system under study is varied over at least the tempera- tu . . . he range which would occur during a measurement run. While the 127 temperature variation is occurring, a TDA routine such as CBTSLS, described previously, is used to monitor both conductance and tempera- ture. The raw data from such a run may be plotted as conductance vs. thermistor response (EIN) by the CCLTLF program which will not be described here. Such raw data is shown in Figure 36 a, b, and c for three systems in dimethyl sulfoxide (DMSO); 4 x 10'42/1 luminol (36a), l0'3fl potassium tertiary butoxide (36b), and the products of the reac- tion between these species (36c) which will be discussed in Chapter 8. These curves cover a temperature range of 5.4 °C, from 29.9 to 24.5 °C. They have been included in this manuscript because they demonstrate several interesting qualitative aspects of the work done with temperature- conductance profiles which will be discussed later in this chapter. It can be seen that the temperature change which occurred during the runs in which the data of Figure 36 were obtained was, itself, non- Iilnear. This is evidenced by the accumulation of an increased density 01’ (data points near the 24.5°C region of the curve (lower right). Iller cause of this effect is the way in which the data was generated, by iiirst heating the solutions in the conductivity cell and then plung- IDSJ the cell into a constant temperature bath at 24.5°C when the data accIuisition began. The cooling was, of course, much faster in the early Sta962's of the run. In general, it is common for temperature to change "OFF-Lnniformally in an experiment. This presents an interesting problem 1" CHJi~ve fitting which is the method employed to construct temperature co"""ec:tion parameters for the conductance of a particular system. If there are more points 1'0 one section of a curve than another, the method of least squares will effectively weight that portion of the c Urve most heavily at the expense of the rest of the curve (unless the 128 0 6. b O ‘5. o. '\ o.‘ S . -- (a) F" . o, .u- .o". p. O. c. s 0,3,. . s’.‘ ' ~ ’17-, o . . 0". .‘A L (b) o .‘f. - . - a. 0’ O a a ’ 0" . . 0“ . O 9 I. '. '1. 0" J . - (C) {5‘ ‘P‘.. . O M. .:\ "‘o o ' w‘~ ." z, 0.. 0'." O.” 0‘. .\0. 4‘ .\ \ O ..J.\ \k N 1 1 4 A 1 L \r 1w J Figure 36. Thermistor Response Termistor Response-Conductance Profiles for (a) Luminol, (b) Potassium Tertiary Butoxide, (c) Products of Reaction of (a) and (b) all in DMSO. 129 curve is perfectly linear). There are two possible solutions to the problem. One is to program the computer to store measured temperature-conductance data only when the temperature is at a certain value, set up by the initial measure- ment and a running increment. This method leaves some doubt as to the length of a particular run, what increment to set up, how many points to take, etc. It has the advantage of automatically constructing a data set with points linear in thermistor response. The other approach was chosen for the computerized conductance system. It involves arranging the raw data set in monotonically de- creasing order with respect to thermistor response. Intervals within the array were created by software. Each interval contains a single thermistor response-conductance point which would be the average of all measured points which fell within that interval. The program which performs this "array arranging", CBTALR, is flowcharted in Figure 37 and is listed in the Appendix. (It should be noted, to avoid confusion for the reader who is also rteferring to the CBTALR Appendix listing, that the flowchart in Figure 37 is considerably simplified. Many of the operations of CBTALR which éir“e explained in FORTRAN terms in the flowchart are actually manipula- tions which are done in assembly language in CBTALR for reasons of Speed and necessity. These include magnitude comparisons, data switches, array addressing, averaging of thermistor response measurements, and $0 forth). (CBTALR begins by requesting the first block on the DECTAPE where the data from the TDA routine was written, reads the file, and sets up U"’ necessary software pointers. The raw data array must consist of a 1:3() (I) '55 HITHIN ? N0 Figure 37. CBTALR, Array Arranger, Program Flowchart. (Simplified) H53 COT is: 1 0' C0 31‘ an ‘n'h in 131 l2 bit conductance word, the 12 bit division remainder, the parameter word, and the l2 bit, averaged thermistor response word. After CBTALR has completed the array arranging process, the data in the array will consist of the 3-word FORTRAN floating point values of the conductance (scale change corrected) and the l2 bit, averaged thermistor response words. The array arranger begins ordering the array in decreasing value of thermistor reSponse by comparing the first temperature measure- ment (T(l) in Figure 37) with all other temperature elements in the array. If a thermistor response is encountered which is larger than T(l), the two measurements are switched in location along with their corres- ponding conductance values. In this way, at the end of the first pass through the array, the largest thermistor response is in T(l); the other values are in indeterminate order. The second pass through the array compares all remaining values to T(2) so that at the end of this pass the second largest thermistor response is in T(2) and its corresponding conductance value is in 6(2). The process is repeated until the entire array is ordered in decreasing thermistor response. The array arranger next begins to scan the array to determine what the largest interval between thermistor response points (X in Figure 37) in the data set is. Once it has determined this interval it sets up the first boundary in which to average the data by starting at T(l) and going to the value of T given by T(l) - X. It zeros SUMl and SUM2 which will contain the running summations of thermistor response data and conductance data respectively, within a given interval. CBTALR then checks each thermistor response word to see if it is within the current interval. It continues to sum these T and G values until a thermistor response value is encountered which is outside this interval. 132 At this point the sums are divided by the number of measurements totaled (Y in Figure 37). The averages are stored back in the array beginning at the location which had corresponded to the first element and continuing to as many locations as needed (given by T(J) and 0(0) in Figure 37). CBTALR then moves the boundary downward by X, resets SUMl and SUM 2 to zero, and repeats the process beginning with the first point to fall outside the previous interval. When the arranging is complete, there are J points left in the array. Only one point may exist per interval, thus eliminating the apparent "weighting" of points toward one region of the curve. Some points will, of course, be more precise than others. The computer outputs the chosen interval and number of final points in the array, and the new array is stored on tape 16 blocks beyond the raw data array. Arrangement of a 500 point array requires approximately 145 seconds. The effect of CBTALR on the data of Figure 36 is shown in Figure 38 a, b, and c. The shape of the curves is unaltered. Some Of the noise has been reduced by averaging of discrete points. The data are Properly suited for fitting by the least squares technique. The data of Figure 38 were plotted by CCLALF. This is the second analysis program in the temperature-conductance coefficient determining routines which use the array arranger. A series of programs was re- quired due to limited core space in the POP-8/I. CCLALF is flowcharted in Figure 39 and listed in the Appendix. CCLALF reads the arranged data from tape, outputs it on the line printer'if the operator desires, and plots the data on the X-Y plotter and/or the display scope. Plotting options include plotting axes or the data set. The Operator may request that any continuous group of points within the data set be plotted. mnz>qncczon 133 "-.(b) [\4 1 L 1 1 1 1 1 1 #4 Thermistor Response F. .ISJIJre 38. Array-Arranged Data From Figure (6-3) (a) Luminol. (b) Potassium Tertiary Butoxide, (c) Products of Reaction of (a) and (b) all in DMSO. 134 l PLOT PLOT AXES CORRECTED Ragg¥¥ozo NO ON x-y YES PLOTTER 7 F1'sure 39. CCLALF Program Flowchart. (Simplified) 135 In plotting, the program searches the array for the maximum and minimum conductance and thermistor response for that group of points and uses this information to automatically scale the data for output to the plotter and scope. If the operator does not require hard copy of the data, he may fast point-plot it in a second or two on the display scope. Point-plotting on the X—Y plotter requires as much as 20 minutes for 500 points. When the operator selects the FIT option, he may fit any con- tinuous block of points within the data set. The maximum and minimum conductance and temperature are calculated for the selected group of points and they are plotted. The program then stores, in the indication portion of the data array, the values of the first and last points to be fitted and the calculated plotter scaling parameters. The entire array is retransferred to tape with these new indicators included. The Operator may, at this time, select any of five fitting programs which generate least squares fits to functions of the following forms: aT + b aT + bT1/2 + c aTz + bT + c ai3 + hi2 + cT + d C) C) G) G3 G) N ai5 + bi4 + cT3 + di2 + eT + f The general form of these fitting programs appears in the flow- chart of Figure 40. The cubic fitting program, CDTALC, is listed in the Appendix. The chosen fitting program reads the array written onto téitlee by CBTALR and CCLALF, initializes software pointers, and proceeds 136 INITIALIZE I DICATORS SETI= #frOALCULATE G( I) AND T(l)] ADD TO SUWIATIONS ALL N0 VALUES sums-:0 2 YES [SET up DETERMINANT l I SIORE COEFFICIENTS ON TAEELWITH DATA I IRETURN TO MONITOR I FIgure 40. Curve Fitting Program Flowchart. to cal artic 'Im 137 to calculate the summations required for the least squares fit to the particular order of equation chosen. Once the summations are complete, the program sets up the determinant to be solved for the coefficients. a, b, c, etc. of the chosen fitting equation. The determinant is solved by a pivotal condensation (47), to help reduce roundoff errors, for all functional forms except the linear function which is solved directly from the least squares formulae for slope (a) and intercept (b). Once the determinant is solved, the equation giving conductance as a func- tion of thermistor response is printed on the teletype. The coefficients of thermistor response are stored on tape within the data array to enable the operator to obtain comparison of the real data and the fitted curve. A 100 point cubic fit requires about 9 seconds of POP-8 run time. The fitted data, real data, and residuals may be listed on the line printer by CELTLR (not included in the Appendix). The fitted curve from any function may be plotted by CFPTLR. Any function with only integer powers of T may also be plotted by the faster CFPTLI, flow- charted in Figure 41 and listed in the Appendix. CFPTLI reads the T—G coefficients determined by the fitting routine and sets up the plotter scaling parameters determined by CCLALF. It calculates one five- hundreth of the measured thermistor response range for that data set and plots the first point as the conductance calculated from the fitted equation vs. the smallest thermistor response measured. It then proceeds to plot the entire measured thermistor response range according to the fitted equation by incrementing T by 1/500 for each successive point. It plots straight lines between points, producing a continuous curve. Pf the plot of raw data generated by CCLALF or CCLTLF has been saved on ”We scope or X-Y plotter, it will, by plotting the fitted curve on the 138 [SET UP SCALING PARAMETERS AND COEEFICIENTS I I E NGE CALCULATE FIRST P01NT lilfilIilfiilflmlflflll IflHHEifliflfllflfllflifiiliillflflflfil no ALL DONE ? YES [RETURN TO MONITQR I Figure 41. CFPTLI Program Flowchart. 139 same scale on top Of the raw data, allow the user to qualitatively examine the fit. When the entire curve is plotted, the program ter- minates and returns computer control to the Keyboard Monitor. It should be noted here that there is an entire series of T-G data analysis routines and fitting programs which run without using the array arranger. These routines were written for the computerized con- ductance system prior to the implementation of the array arranger. They are essentially analogs of the programs discussed above. They are still useful where temperature variation can be relatively linear although they are seldom used due to the simplicity and speed with which the array arranged data can be manipulated. C. CONDUCTANCE DATA ENHANCEMENT THROUGH TEMPERATURE VARIATION CORRECTION Figure 42 shows a raw data set with its corresponding fitted curve as discussed above. The raw data displayed in Figure 42 is the data in the form in which it was acquired by CBTSLS, before array arranging. The fitted curve was calculated for the arranged data. In this case the data was nearly linear, but sufficiently curved so that all terms in the cubic fitting equation were of nearly equal size when the magni- tude of T is considered. This particular fit was used to correct the data of Figure 43 for temperature variation. This run was performed to demonstrate the power of this temperature correction technique. The TDA run involved varying the temperature of a dilute sulfuric acid solution 15 °C while no bulk ionic change occurred. Thus, the only 3 conductance change (about 2 x 10' 0'1) which takes place is that due to the temperature change of the system. The data of Figure 43 are 140 AmieumPFV .AUomm . mmv Empmam smug: . uwu< upcacpzm um>Pommecmga a cow mmcmcu mcapmsmaeme comp a cow pee ownau use mpwcosa mucmpuavcou . assumcmaewe .Ne mesmem 8.3 3833. Sergei 9.8 a a s . a a e S x S: T N- L 72 x engage + e Nu: x 338:5- 1 Ne m-o_ x Nomemmme.o + me e-o, x eooemmme.- u u . . _-e ~-o_ x m_m._ OUOQ UOUOULLOU 0L3.» OLUQEOF 11M .:_...I,\|\.U\ Kai‘s!» 1\ 141 Am-eea__v Aeom~-wmv emceeo oeeoc teem oe new; oom— empmsm capo: - v_u< Ovccham a we mcwpoou soc opwcosa we?» . mucepuaccou .me acumen Aummv weep omN com omp cop om o aw s a u a a. a J_ u n1 Pic nopxepm.P I spec umpuwccou u 38 267/ //:1. f / sumo umuumssou assuasoaeme {1F LIL I pi UOZQDUH owp zueeum Paame ace - ------ «85.5 . H can P LO :5 N >m~+ .l no r- -'--"'-|-|-"""'L -‘I'-I--"-'-L|--'---"- pares ------------- -----------1 xp P--------'--1 I xue >m+ mamm 155 for supplying DS-IOP signals (08 36, IOP, l, 2, and 4) to external devices used with the computerized conductance system. The cell which was used for most of the titrations investigated with the computerized conductance system was constructed from a 100 ml three-neck, round-bottom flash with two platinum wire electrodes mounted in the side, facing each other end to end. There are no platinum discs on the ends of the wires and the wires themselves are not platinized. The side of the flask was idented above the electrodes to provide a "shelf" which shields the electrodes from waves formed on the surface of the solution by the action of stirring. Stirring is accomplished by a 3/4" glass propeller placed in the center of the cell through the central opening of the flask and connected to an over- head stirrer. A glass bushing is provided around the propeller shaft to hold it in place. The burette delivery tip and the thermistor probe are each mounted in ground-glass joint plugs which can be inserted into the other two openings of the cell and held firmly in place. Most of the cell is immersed in a constant temperature bath. The measured cell constant, k, is defined as K II KR where K is the specific conductance of a particular solution in '1 and R is the measured resistance. The constant, k, for this cell was found to be l.889:.002 cm". 9'] cm 156 B. TITRATION DATA ANALYSIS AND DISPLAY SOFTWARE Once the computer has completed a titration and written the data onto DECTAPE, the operator will call CCLMLT, the first of two analysis routines for titration data. Program CCLMLT is flowcharted in Figure 48 and listed in the Appendix. It begins by asking the operator for the first block on tape which contains the data to be analyzed and then reads that data. It determines, from the parameters written into the data set, whether or not temperature and double precision data were taken. If it finds that temperature data were taken, the computer asks the operator to input the coefficients from the temperature-conductance profile fit for that system. (The operator may input zeros for these coefficients if the temperature-conductance run has not yet been performed). Finally, the computer asks for the initial volume in the cell before the titration began so that the data may be corrected for dilution effects, and whether or not the operator wishes to have the data listed on the line printer. If a line printer listing is to be made, the computer will set up the line printer and data table headings. A set of maximum and minimum pointers, for all types of raw and corrected data, equal to the first points in the data set, will be established and the analysis begun. The sampled voltage, ES, the raw conductance measured, 0(M), the conductance corrected for scale changes, 0(C), and the dilution corrected conductance, 0(0), will all be calculated. The quantity 0(0) is calculated according to 6(0) = (vi + vtit)(G(c))/Vi 157 MIAMI I.i IUAII'I‘JL'JI'IW12.11131: IUJ'IU'JI i DATA '55 um i no IJI'JJHIHNNJI'I'TW Hall'.‘ll.‘IT-]Il'flI'7: - ; :1! . .11 IV HOMBRE)“ ECWDKHGIWEI wax on YES MIN ? no .131" ’ DIS-l [(M- E '1 '1‘ . DHIN GEM-I w, ,v v" 'v H. . V" "E" 0T TT 1 T I. , i . '0 mm. as. am. am “:13" G T AND G D-T 7 NO YES ‘IIIIIITI’IJIUIH ‘4 OUTPUT MAX ANDTMIN GgT) l:Nl'u‘llbllilflifllz Figure 48. CCLMLT Program Flowchart. (Simplified) 158 where V1 is the initial volume (input by the operator) and vtit is the volume of titrant added. The computer will compare each calculated conductance to the corresponding maximum and minimum in the set and reset these limits if a new maximum or minimum is encountered. If temperature data have been taken, the conductance corrected for temperature fluctuations 0(T), will be calculated from: g n-1 n-1 n-2 n-2 0(T) 6(0) + (An)(Ts - T ) + (An_1)(TS - T ) + ... where An is the coefficient of the n-l power of the thermistor response, T, from the fitted equation (input by the operator) and Ts is a standard reference thermistor response, measured at the beginning of a run, before any temperature change has occurred. Conductance corrected for dilution and temperature, 0(D-T) will be calculated from an expression identical to that for 0(0) with 0(T) substituted for 0(C). Maxima and minima are also calculated for the 0(T) and 0(D-T) data sets. If the data are to be listed on the line printer, the computer will output the conductances for each point, 0(M), 0(C). 0(0), 0(T) and 0(D-T) as they are calculated along with the point number, the volume of titrant added to that point, and the sampled voltage, Es‘ Thus, in the few minutes following the actual titration, the computerized conductance system has completed calculations which would require many man-hours Of tedious work if the data were taken by conventional means and treated in a conventional manner. Finally, when all the data have been scanned, the maximum and minimum values for all conductances calculated are listed on the teletype for use in setting up the scope 159 and X-Y plotter boundaries for data display. Computer control returns to the Keyboard Monitor. The plotting routine for titration analysis, CDPMLT, is flow- charted in Figure 49 and listed in the Appendix. It begins by requesting the first tape block to read, reads the file, requests correction co- efficients if temperature data were taken, requests the initial volume, and allows the operator to plot axes, 0(M), 0(C). 0(0), 0(T), or 0(T-D) vs. volume of titrant, or return to monitor. When the operator selects an option to plot one of the calculated conductances, the program allows him to select the points to be plotted and the upper and lower conductances to be the limits of the plot. The computer then calculates the chosen conductance data point-by-point, scales it for plotting, and plots it on the X-Y plotter and/or display scope, drawing straight lines between points. When the plot has been completed, the routine returns to await the selection of the next Option. It has not yet been necessary to provide a fitting routine to calculate the endpoint of a conductometric titration analyzed by these programs since the endpoints for most of the systems investigated to date have been quite distinct. Furthermore, the conductance change during these titrations is being continuously monitored within the quantization level provided by the titrator (0.2 ml/ZO ml). Having 100 connected points to define a curve largely eliminates the uncer- tainty in the endpoint which some workers have experienced using conductance detection. These persons only obtained a few measurements on either side of the endpoint and draw straight lines through them, taking their intersection as the true endpoint. In fact, one of the most interesting phenomena presented in this chapter (section 0) 160 Aeoec__aeemc .2223: 59.595 52.26 .me oceaea 161 would have been completely unobserved if continuous monitoring of the titrations had not been employed. The use of a TDA routine for titra- tion data acquisition, an automatic titrator under computer control, and computer evaluation of data made all of the titrations presented here easy to perform and analyze for the chemical information they presented. This system makes the performance of sophisticated titra- tions, and the analysis and display of the data, possible for even the novice experimenter. C. EARLY STUDIES WITH THE COMPUTERIZED CONDUCTANCE SYSTEM: PRECIPITATION TITRATION OF Ag+ WITH KCL The earliest chemical measurements which were made with the computerized conductance system were determinations of Ag+ by titra- tion with KCL. These measurements were performed when the titration software and hardware were being debugged. Figure 50 shows a typical early titration curve for 100 mls of 0.0011flAgN03 titrated with 0.010 M KCL. A sharp break can be seen at the end point. The overall conduc- '59'], which corresponds to tance change for the titration was 5.20 x 10 a 44.3% relative conductance change. The results from a number of titrations of this system were precise to within 1.23%. The instrument was required to change scale three times during the latter portion of the titration as evidenced by the three discontinuities which appear on the plot in this region. This titration was performed before pro- visions had been made for software correction at scale changes and overlap of adjacent scales. These discontinuities indicated the need for such corrections which were later implemented. The computerized 162 rise .§_ m 08.0.5; +2 to 83: :8: to 5:5: 35623328 .8 2%: o.o~ Anacoeogocn FE N.ov covc< pox $0 .mp5 o.o TAIHVTT .1 . 14 p 1 . d . 3-2522 sumo zem aceoa emcego mFeom .- k. QOZQDUI—(ZULU 1 73-25;: 163 conductance system did, however, demonstrate its ability to measure conductance changes in chemical systems where precipitation occurred. No effects due to silver chloride formation on the electrodes were Observed. Concentrations of Ag+ as low as 10"4 M were successfully titrated. D. SOME OBSERVATIONS CONCERNING THE END POINT PHENOMENON ASSOCIATED WITH CERTAIN EDTA TITRATIONS During the early chemical measurements with the computerized conductance system, what was assumed to be another "simple" titration was examined, largely to determine titration sensitivity in the pres- ence of a strong background electrolyte. This system was the titra- tion of Ca++ with EDTA in ammonia ammonium chloride buffer at pH 10. These titrations were performed at four separate times, in June and August of 1973, and in January and March of 1974. The first two studies were performed to assess the performance of the computerized conductance system in buffered solutions, where there is a large conductance background due to the buffer ions. The later two studies were attempts to investigate and duplicate the end point anomaly Observed during the August, 1973 study. The initial set of titrations, performed in June, 1973, were observed to be normal conductometric titrations with normal end points. During this study, Ca” concentrations as low as 1.25 x 10'4 M were determined by titration with EDTA. The buffer present in the calcium solution had an ionic strength of about 0.05. The end points were distinct and the accuracy, 0.625%, was limited by the solutions prepared. The precision was excellent, duplicate experiments 164 agreeing to within 0.020%. Further titration studies were, however,' delayed for about two months while the temperature monitor circuit and software were assembled and integrated into the computerized con- ductance system. In August, 1973, the chemical tests were resumed with further investigation of the Ca++ - EDTA titration among the experiments to be performed. The results of this series of Ca++ - EDTA titrations were totally different from those Observed in June. Instead of the normal titration curves previously observed, the region around the endpoint of these new titration curves was significantly altered. One of these curves appears in Figure 51. Two breaks near the endpoint occur where only one existed previously. The instrument was thoroughly checked and the cell thoroughly cleaned but the phenomenon failed to disappear during the entire five days when these runs were made. The phenomenon was found not to be a function of buffer strength, concentrations of Ca++, concentration of EDTA, any tested ratio of these concentrations, or any particular instrumental settings. It was found that the true Ca++ endpoint was obtained by graphically finding the point half way between both breaks as shown in Figure 51. It was discovered, at the time, that a number of previous workers ++ ++ , Cu had noted similar effects for Ca++, Zn , and other divalent metals with EDTA titrations in buffered solutions. The earliest of these was in the pioneering work of Hall, Gibson, Wilkinson, and Philips (51) who were the first to use conductometric methods for determination of endpoints in EDTA titrations. They noted curvature in most of their titration curves which could not be explained by dilution effects. This curvature was not, however, as dramatic as 165 .. o.o_ Ia ea soccem +ezz - mzz er (ecu.mcmoo.o new: ++eu z~oeo.o co ape CF to eoeoeeoee ..m mesmea o.o~ Amoeoeosoee .5 N.ov eoee< HOCDZOD 7. 260x10‘40' ‘ 8.821x10'4 '1 MOZ>HOCCZOO '.473x10‘40" _ 172 ._ a 31874-1 Dilution Corrected _. Data .- xix” ,_ 1 1 1 1 1 1 1 1 ._1 1 ‘- b Dilution Corrected 31974- Data ,. l L l J l l l I J J Figure 53. a) Conducticator effect for 4xlO'4M CgH Titrated with 0.002 M EDTA in the presence of 2x10‘ M K3Fe(CN)5 and 0.01 M NH3 NH; Buffer at pH 10, (b) Same as (a) but no KFe(CN)5 Present, Cell has been Scrubbed three Times w th HF and HCl. 173 were prepared in previously unused glassware. This is reflected in Figure 53b which represents the titration made following the above cleaning. The conducticator effect now appeared to be a surface phenomenon within the titration cell, either on the glass or the platinum electrodes. The effect was finally eliminated by cleaning the cell with con- centrated nitric acid. It was then expected that the effect could be recreated by addition of fresh Fe(CN)6'3 as in the first experiment. However, the effect could not be recreated although the exact experi- mental conditions and the conditions preceeding them were carefully duplicated. Treating the surface of the cell with HF to remove the outer monolayer of glass, exposing a fresh surface, produced no effect. Treating the cell with concentrated NaOH to restore the Na+ balance on the glass (since an ion exchange with the glass surface seemed a reasonable mechanism for the conducticator effect) had no effect. Finally, having the cell re-anealed had no effect. Since other studies with the computerized conductance system had been undertaken at this time, and because of the lack of further results, the investigation was halted. Several conclusions can be made at this time which might be of use to later workers investigating this effect. They are: l) The conducticator phenomenon is real, having been observed by a number of workers with different instruments under different environments. 2) The effect could prove to be valuable in determination of lower concentrations of analyte by conductance than previously pos- sible. However, if the effect is due to complex equilibrium 174 between the conducticator ligand and conducticator-EDTA the chance Of finding such a species which could be used in numerous systems, where complexation constants can be quite different, seems to be rather small. 3) The effect is due to an interfering species and not to the formation of an EDTA-buffer complex. 4) The conducticator phenomenon appears most likely to be a surface effect, possibly an ion-exchange on the glass surface of the cell, rather than an effect resulting from the presence of a species in solution, since careful cleaning of the cell failed to remove the effect. 5) CN' in some form seems to be a species which will trigger such an effect. 6) The equivalent conductance of the ions released to the solu- tion when the phenomenon occurred, per mole of EDTA added (calculated from data in Figure 53b was 579.90'] cmZ/mole EDTA. This almost exactly corresponds to three OH' species (total 1Slcmz). equivalent conductance, 576.0 ' E. DETERMINATION OF SMALL AMOUNTS OF NaOH IN THE PRESENCE OF LARGE QUANTITIES OF SODIUM PHENOLATE The author became aware of an analysis problem which a chemical company was experiencing while attempting to determine small amounts (Pom .uom.¢m an voumumOELmsh ppou :ox z moopo.o saw: Uwum uwpocuzn z mpoo.o mo mFE 0.0m *0 cowumguwh uvgumeouoaficou .om mgsmwm 0.0N tocum :02 $0 mpE o q _ . _ _ _ _ m 3% I. empumeeou cos: _. E .l pic miop x mumm.o 182 phenomenon associated with each portion of the curve (A, B, and C in Figure 56 is given below: Region A: The increase in conductance in this region is due to the reaction: K++OH' _ + PH2---* PH + K + H20 where two ions not previously present (PH' and K+) are being formed. Region B: The conductance in this region of the curve decreases very slightly. The increase in conductance due to the addition of K+ (with OH' reacting with the second phthalic proton to yield H20) is not seen. In addition, the product formed is of slightly lower conductivity than the PH' Species. The reaction which is occurring here should correspond to: _ K++OH' PH ———" PK- + H20 where APK' is slightly less than APH” Region C: The conductance increase here is due to addition of K+ and OH'. Notice that the conductivity increase is not as rapid as that for the fOrmation of K+ and PH' in Region A. Two possible conclusions can be drawn. Either AOH' < APH‘ or some PK2 fonms when excess K+ is present after the second endpoint. It is quite possible that the first conclusion may be correct due tolfigher salvation of OH' by the bulky DMSO molecules present be- cause of its higher charge-to-mass ratio with respect to PH'. 183 This titration, along with the others presented in this chapter, demonstrates the versatility of the software and hardware in the computerized conductance system, and the power of its instrumental sensitivity and automatic data analysis, for applications in titration studies. These titration abilities may now be implemented by other workers, who desire to use conductometric monitoring techniques, with a minimum of effort and the anticipated maximum data attainable by the conductance method. CHAPTER 8 APPLICATION OF THE COMPUTERIZED CONDUCTANCE SYSTEM TO KINETIC STUDIES: PRELIMINARY INVESTIGATION OF THE LUMINOL REACTION Acknowledgment: All of the actual experimental work and much of the implementation of specialized hardware presented in this chapter was done in conjunction with Timothy A. Nieman to whom the author wishes to express his gratitude. A. SPECIFIC SOFTWARE FOR KINETIC EXPERIMENTS One of the greatest differences between the computerized conductance system and conventional conductance instruments is the speed with which data may be acquired and analyzed. At its maximum rate the system can acquire over 33,000 measurements of conductance per second. It was shown in Chapter 5 that these measurements were of good precision and low noise level. If the timing of an experiment permits, ensemble averaging can increase this precision and lower the noise level accordingly. Thus, the system is ideally suited for and, indeed, was intended to be used in, kinetic studies of slow to moderately fast reactions in which an overall ionic change is occurring. In order to permit use of the system in reaction mechanism and rate studies, some modification of the basic software was required. As a result, there are currently two sequences of kinetic study programs in use in the system. One of these sequences performs acquisition and analysis of conductance and temperature data (the CBTMLS sequence). The other sequence (the CBTDCS sequence) is also capable of acquiring and analyzing data from another data source, in addition to acquisition 184 185 and expanded analysis of temperature and conductance data. The program CBTMLS is a simple modification of the basic acquisi- tion program, CBTSLS, discussed in Chapter 4. It contains stopped- flow flag check instructions to permit data acquisition triggering from the stopped-flow apparatus employed in most of the kinetic studies performed to this time. Analysis of the data acquired by CBTMLS is accomplished by CCLMLS. Program CCLMLS is identical to the CCTMLT program used in titration analysis, but does not contain provisions for dilution correction of data. The plotting routine in this sequence, CDPMLS, is identical to CDPMLT used for titration curve plotting with the exception that it lacks provisions for plotting dilution corrected data. The other kinetic program sequence contains two computerized conductance system programs which require 12 K of memory for opera- tion. Program CBTDCS is a modification of CBTSLS which is capable of acquiring conductance, temperature, auxiliary data, and a baseline for the auxiliary data. It contains additional storage space which enables it to acquire up to 500 points from spectroscopic, electrochemical, or other data sources. It has instructions to trigger the auxiliary A/D converter on the same signal (OS 33, IOP 4) as the bipolar pulsing trigger. It uses 05 36, IOP 2 for a flag check of the stopped-flow apparatus and DS 36, IOP 4 for the data transfer. Data are analyzed and plotted by the 12 K CCLDCS program. This program is an extensive nodification and combination of CCLMLS and CDPMLS. In addition to listing the calculated data on the line printer, calculating maxima and minima, and slow plotting raw, scale change corrected and tempera- ture corrected conductances, it will fast plot all of these values on 186 :4. ~ ' no wq l 3' «r6. ‘* “- ‘ ' " '3’1’.1f\9:.‘ .1 (st. .$:‘€ ‘35: 9, ~—..‘\ . .t 1:1 , . \ "M - 1. ‘ Figure 57. Photograph of the Early Hacker Stopped-Flow Apparatus. 187 the display scope, list them on the lineprinter, and slow or fast plot thermistor response, auxiliary data, and the integral of the auxiliary data (baseline corrected). The fast plotting option permits an operator to observe his data on the display scope within about a minute of the completion of a run. The only waiting period necessary between completion of data acquisition and plotting (about 45 seconds for a 500 point data set) is that required for the computer to calculate the set of maxima and minima used for scaling of the data on the scope. Plotting of data requires two-to-ten seconds for 500 points. If an Operator desires a hard copy, data may be slow-plotted on the X-Y plotter in the normal manner. ‘ B. SPECIFIC HARDWARE FOR KINETIC EXPERIMENTS The stopped-flow apparatus which was used for the studies to be discussed later in this chapter is shown in the photograph of Figure It is an early version of a custom stopped-flow device built by It utilizes compressed air-driven syringes It has been 57. the Hacker Machine Company. which inject a total volume of about 1.2 ml per trigger. fitted with a combination conductance-temperature-spectroscopy observa- The cell has quartz windows at either end to permit tion cell. The flow pathway in the cell was monitoring of spectroscopic data. bored through four platinum disc electrodes which are sandwiched in The electrodes are spaced such that, by choosing appropriate Kel-F. pairs of electrodes, the distance between conductance monitoring points can be selected as 2, 3, 5, 7, 8, or 10 cm. In addition, all four electrodes could be used at one time with the four-lead computerized conductance system. A thermistor is imbedded in the cell wall, touching 188 the flow stream, to permit the cell temperature to be monitored. TO minimize temperature changes, provisions are made for thermostating the cell directly or in combination with the rest of the apparatus. At the time of its use in this study, this apparatus had three problems associated with it which adversely affected its use in kinetic studies. The first was a rather long dead time of about 40 mseconds which prevented its use in initial rate studies of fast reactions. The second problem was that more heating was generated by mixing than would normally be desirable (the exact temperature change depends on the viscosity of the solvent, but is on the order of 0.1 to 0.3°C even with thermostating). Finally, the solution was in electrical contact with the chassis of the instrument at the outlet of the cell. For solutions of high conductance, this causes the measured conductance signal to be intolerably noisy. The noise can only be eliminated by unplugging the apparatus which, of course, rendered it useless since the activation switches are electrical. Grounding the two electrodes closest to the outlet (the left electrodes in Figure 57) did diminish the noise somewhat. Nevertheless, in its condition at the time, the device could not be used for kinetic studies by conductance where the measured resistance between the two right electrodes is less than 20 K0. It is hoped that fitting the newer Hacker stopped flow apparatus with a conductance cell will enable these problems to be overcome. However, for the studies presented here, which involved a relatively slow reaction and solutions of low conductivity, the earlier apparatus proved suitable. In many of the experiments which were performed with the luminol reaction, it was desirable to obtain light data from the ”...-1“ 189 chemiluminescence (CL) of the reaction as well as conductance and temperature data. To implement this measurement, an auxiliary data acquisition system was added to the computerized conductance system as shown in Figure 58. The CL from the reaction was followed by positioning a Heath EU-70l-30 photomultiplier module containing a 1P28A Since only the photomultiplier tube at the observation cell window. The total light intensity was followed, no monochromator was needed. current output of the photomultiplier module was amplified and converted to a voltage signal by a Heath EU-70l-3l Photometric Readout Module. A supplimentary amplifier, A6 (Analog Devices 1428) was used to amplify the Photometric Readout Module output to levels near full scale for a 0 to +10 volt A/D converter. The output of amplifier A6 is sampled by an Intronics FS 201 sample-and-hold module. When the computer outputs a convert command to the A/D converter, the STATUS output of the A/D converter goes high, causing the sample-and-hold module to hold the data until conversion is complete (when STATUS The A/D converter, an Analog Devices ADC-12QU, was returns to low). It can perform a 12 bit set in the 0 to +10 volt conversion range. conversion in 15 nseconds. Since it is being triggered by the same signal which initiates the bipolar pulsing sequence, which requires a minimum of 30 nseconds, no provisions for a flag for this circuit were necessary. The MD output is transferred to the accumulator :hrough a Heath EU-BOO-JL gated driver, in the same manner as the onductance data discussed in Chapter 3. A flag is provided for triggering the timed data acquisition aquence by the stopped-flow apparatus itself. The stopped-flow evice contains a relay which closes when the drive syringes close 190 .couecoz memo mucoumo=e534 .mm mcamve 304m hum. I e on e mm --. 1330.5 mom we no a mmxuzou can: oflmmmummmma u o M o :2 EN 111 3%,; 2m. :5: 191 and which can be used to switch a signal to some external circuitry. The card containing the logic gates of the flag circuit of Figure 58 was placed in a peripheral control slot available in the digital circuit compartment of the conductance instrument module. When the drive syringes close, the relay closes, which grounds the input to the first NAND gate and places a “l" at the input to the second NAND gate. A DS-IOP signal will cause the output of this gate to go low, which causes a skip and allows the computer to begin data acquisition. It was necessary to combine this inverse function (S—KP) by a logical AND with the corresponding output of the conductance flag circuit in order that one of the I/O cards already within the instrument module could be utilized with the stopped-flow flag. Later workers using the auxiliary acquisition hardware of the computerized conductance system should be able to acquire data other than simple light data with only minor modifications to the conversion system itself. C. PREVIOUS INVESTIGATION OF THE LUMINOL REACTION The chemiluminescence reaction of luminol with base was first observed by Albrecht (55) in 1928. Albrecht studied the aqueous reaction of luminol in alkaline solution with hydrogen peroxide in :he presence of a catalyst such as potassium ferricyanide. The reaction, n both aqueous and aprotic solvent systems has the highest quantum ield (about 5%) of any non-biological CL reaction studied to date. lthough many workers have investigated the CL and fluorescence Spectra this reaction system since Albrecht's initial study of it, little rk has been done to define the complete reaction mechanism. 192 White and his co-workers (56,57) have shown that the overall reaction appears to be: ° 11 11 c - C‘N-H 20H' 02 in DMSO) \0 _- + - + 2H 0 + N CI,N H 11202 plus 0/0 2 2 I ll NH2 0 metaloin NH2 0 2 3-AMIN0PHTHALATE-2, 3-AP'2 LUMINOL. LH2 They studied the reaction in DMSO containing 30 mole percent water The products of the reactions in aqueous, semi -aqueous, and non-aqueous media appear to be the same although the reaction in DMSO is somewhat simplified. In DMSO the reaction takes place with no stronger oxidizing agent than molecular oxygen present; no metal catalyst is required in these solutions at all. White et al. proposed the only complete reaction mechanism which has been suggested to this time for the luminol CL reaction: .‘1’ 0E 00:2: C’ N-H sH—o—' N-H ($53 C/ (slow) ”“2 :1) (fast) 3 L112 LH' 1‘2 11* 9 (I E c 635' it \o .1. 0 ex.- (FD/N- (fast) c/°_ C/o- H N I 2 2 0 H i ll 2 o 0 3-AP -2 *-2 L02 3-AP 193 They felt that the initial proton transfer occurred in two steps since, when no excess base was present to remove the second proton, no CL Thus, the LH' species reacts only very slowly or not at 2 in an excited state was occurred. all with 02. They were convinced that 3-AP' the luminescing species since they found that the CL spectrum of the reaction matched the fluorescence of 3-AP which they were also able to isolate as the almost exclusive product of the reaction. The reaction they studied was found to be first order in luminol, base, and 02. The CL from the reaction in DMSO is dampened by the presence of water. McCapra (58) reports the maximum of this CL to be at 425 mu in water and 485 mo in DMSO. Other workers (59,60,61) have investigated the spectral charac- teristics of the emitter in the luminol CL reaction. The result of this work has largely been to support the mechanism of the reaction as proposed by White et al. and to reinforce the belief that 3-AP"2 in an excited state is the emitting species. Drew and Garwood (62) claimed to isolate a compound from the reaction mixture which contained two oxygen atoms bridged between the two nitrogens of luminol. This lead them to believe that the reaction intermediate formed after the reaction between L"2 and 02 might be a bridged species. Other workers have stated their belief in this form of an intermediate, although there is some possibility that the species observed by Drew and Garwood is a simple luminol salt, solvated by peroxide. NO other intermediates iave been proposed. Gorsuch and Hercules (60) have performed the only stopped-flow tudies of the luminol reaction published at this time. The system hey studied was the reaction between luminol and potassium tertiary 194 butoxide (PTB) in DMSO with 0 to 40% water present. They found the decay of 3-AP“'"2 to be very fast and the rate of the CL decay to be determined by either the overall proton transfers, the L"2 reaction with 02, or the rearrangement of L02"2 to form 3AP*'2. They assumed that for low concentrations of luminol and excess base and 02, the proton transfer would be very fast and, thus, not the rate determining step. Since their CL decay curves for low PTB base concentrations became first order after reaching a CL maximum, they were convinced that the rearrangement of LOZ'2 was controlling the reaction rate after the CL maximum. Finally, the rate constant which they proposed for this 1 reaction in pure DMSO was 1.210.3 x 10'1 sec" . Besides the interesting phenomenon associated with the luminol CL, the reaction itself has proven to be a useful analytical tool as indicated by Seitz and Neary (63). The reaction has been used for detection of trace amounts of metal catalysts and oxidants which react with luminol to produce 0.. Concentrations of Co (II) as low as 1041!. 01. (II), Ni (II), Cr (111), Fe (II), and Mn (II) from 10"8 to 10'1011, and OCl', 12, Mn04', and H202 from 10"9 to 1040! are detectable in the luminol CL reaction. Furthermore, these responses are, in most cases, linear over three to four orders of magnitude. Because of its analytical usefulness and the lack of substantiating evidence concerning many of the proposed reaction mechanism steps or 'ates, it was thought that a study of this CL reaction by other than pectroscopic means could provide additional information unavailable 9 previous workers. Thus, the luminol reaction was investigated with 1e computerized conductance system. 195 STOPPED-FLOW STUDY OF THE LUMINOL REACTION BY MONITORING CONDUCTANCE, TEMPERATURE, AND CHEMILUMINESCENCE CHANGES The first investigations of the luminol reaction with the com- iteri zed conductance system were performed with the CBTMLS program equence. The basic aim of these early experiments was to determine f a conductance change which corresponded to the reaction could be 1bserved. Potassium tertiary butoxide (PTB) was the chosen base. 50th luminol and PTB were prepared in pure DMSO. Karl Fisher titration of DMSO showed the water content to be always less than 260 PPM. One of the earliest reaction curves obtained from the conductance measurement is shown in Figure 59. This curve resulted from the reaction of 2.52 x 10'511 luminol with 1.37 x 10'2 PTB in 01150. (The concentra- tion of the base was somewhat difficult to determine since there often appeared to be a residue of base remaining after attempts to dissolve it in DMSO. Furthermore, the PTB-DMSO solution would deteriorate as PTB reacted with oxygen present in solvent over the period of about an hour as also observed by Gorsuch and Hercules (60). Figure 59 shows a rapid and smooth conductance change upon mixing the reactants. 1000 points were taken, each every 0.025 seconds, for this run. When the curve is expanded by plotting only the first 100 points, most of the conductance change can be seen to occur during the first 0.25 seconds. This particular run was performed at room temperature (about 19°C) without thermostating the cell. The results appear encouraging, but it was necessary to prove that the curve was not a result of temperature changes or base or solvent dilution effects on the conductance of the species involved in the reaction. Figure 60 shows the change in conductance of a thermistor in the 196 N beam, oeoaK+ + H8 + Neg? + H8 + L‘Z, where the conductance increase which was being observed was due to the liberation of free K+ and L'2 in the solution where no 8' or K+ was present to contribute to the conductance before reaction. Since little was known about the base strength of KOH in 1:1 DMSO-EtOH (the final solvent system after mixing), the computerized conductance system was used to investigate it. Figure 61 shows the dilution-corrected conduc- tance curve obtained when 0.1003 M KOH in 1:1 DMSO-EtOH is added to 50 mls of 1:1 DMSO-EtOH. (Dilution correction of the data has the effect of making the X-axis on this plot, which actually corresponds to mls. of KOH added, linear in increasing concentration). The concentra- tion range represented here runs from O to 0.01866 M, There is no noticable curvature in the plot at higher concentrations which would occur if KOH were a weak electrolyte. Figure 62 shows a plot of conductance/concentration, 0/C (the equivalent conductance, not 1,2. For a 1:1 electrolyte such as corrected for cell constant) vs. C KOH, the conductance should increase slowly at lower concentrations and intercept the G/C axis if the electrolyte is strong. If the elec- trolyte is weak, the value of G/C will increase sharply as C gets very small due to increased ionization at low concentrations. No such increase is evident on this plot for concentrations as low as 4 x lO’IM, Other such curves were prepared for concentrations as low as 8 x 10'35, No rapid increase in conductance occurred as infinite 201 .uoo.m~ um :oum1omza Pup cw :ox we covpacpcmucoo ucpmmmcucH new: wmceeu mucmpuzccou «my coruocucmucou eo.o ..e mesmea fl mama umpumccou cowuapma is m m _-ee- iopxmom._ QOZQDUE—(ZUM opxocw.¢ 202 AN1eNe_mv .u.o.mN ea :oumiomza Pup ce :og com Accruecucwucouv .m> Au\wv mucmuuzucou pampe>v=om .Nm assure N: ~.o «\Fu c.o 1|| _ — _ _ q A _ d _ o o can: mo.o 203 dilution was approached. One final experiment was performed to assure the workers that KOH was indeed a strong electrolyte. A solution of KOH was prepared and its conductance measured. A quantity of dicyclohexyl-lB-crown-6 sufficient to complex all of the potassium ion present was added. This crown-K+ complex is known to be highly preferred in solution and has approximately the same conductivity as K+. If KOH were a weak electrolyte, there would be a significant increase in conductivity, due to ionization, when the crown ligand was put into the solution. If the KOH were already completely ionized, addition of crown ligand would, at most, cause a slight decrease in conductance by replacing the solvent modules surrounding K+ by the somewhat larger crown ligand. The latter effect was observed experimentally, further indicating KOH to be a strong electrolyte in 1:1 DMSO-EtOH. This disproved the hypothesis that the observed conductance change was due to formation of free K+. Stopped-flow studies of the reaction of phthalic acid and phthalamide with KOH were performed. In these reactions, the only species available to react are protons. In both reactions, an immediate increase in con- ductivity resulted, which could not be followed using this stopped- flow system. In the case of phthalic acid, both proton transfers occurred in less than 100 mseconds. For the single proton transfer from phthalamide (chemically similar to luminol but with no NHZ group on the aromatic ring and only a single nitrogen in the heterocyclic ring), the reaction is complete in at least 100 mseconds also. Thus, it appeared certain that the conductance curve being Observed for the luminol reaction was not due to proton transfer which was found to be 204 very fast for species which were chemically similar to luminol, compared to the rate determining step being monitored. The stopped-flow reaction study performed with phthalamide and KOH indicated that the products of that reaction were more highly conductive than the reactants. This was demonstrated by a definite and immediate conductance increase upon mixing. Since the reaction is given by ,, 1? C \ .1. _ C\ - + -H + K + OH --’ IN + K + H20 ‘11 R. 0 the phthalimide anion must be of higher conductance than OH'. This is possible due to association of the relatively heavy solvent molecules present with 0H" which has a relatively higher charge-to-mass ratio than phthalamide anion. In addition, the charge on OH’ is localized, whereas in phthalamide anion, the charge is very likely delocalized, causing phthalamide anion to be even less affected by solvent interaction. If this is the case, it could be expected that the particular form of 3-AP produced in the luminol reaction will be of higher conductivity than the reactants and that this species gives rise to the observed conductance curves. In order to show that the major portion of the conductance curve for the luminol reaction might correspond to the production of some form of 3-AP, it was desirable to compare the time at which the conduc- tance curve reached its maximum with the time Of disappearance of the 205 CL. Since the CL has been convincingly shown to correspond to the formation of 3-AP*, these two curves should, in time, reflect the same ‘formation. Figure 63 shows the time relationship between the CL curve and the conductance curve obtained for the reaction of luminol with alcoholic KOH. The CL curve reaches its maximum in about one second and decays to near the baseline in about 2.5 seconds. The conductance curve does not level off until almost 15 seconds has elapsed. These curves were typical of the curves obtained for luminol-KOH reactions. Even with temperature correction, the conductance curve required almost 5 seconds to reach its maximum (see Figure 46, Chapter 6). Furthermore, the overall conductance change for the reaction of luminol and KOH was much greater than that for comparable concentrations of luminol and PTB, and much greater than could be explained by the conductivity differences between the reactants and the 3-AP species formed. The cause of this phenomenon became apparent when alcoholic KOH was injected into pure DMSO in the stopped-flow apparatus. A conduc- tivity curve similar to the one presented in Figure 63 was obtained, even when no luminol was present. This indicated that, in addition to the recognized temperature effects, there were long-range solution effects which influenced the conductivity of KOH. Subtracting the curve obtained from the run in which no luminol was present during injection from a run in which a reaction occurred produced the curve shown in Figure 64. It can be seen that the overall conductance change (from 8.8 to 9.8 x 10'60']) is much closer to that observed for comparable concentrations of luminol and PTB. Furthermore, the conductance change levels off in about 7 seconds. Temperature correction 206 Amruemwmmv .Uoo.mm um cmpmumoEcmcp p—mu .IOpm cw :ox.m¢—wmo.o new omzo :e —oc_e:4 zommo.o eo cowpoemm we“ sock umuzuoca m>c=u Jo ecu m>c=u mucmpuzvcou .mo mczmwm Aummv we?» mm cu m_ op m o.o _ _ _ q _ a. _ _ .1. 4 _ . . a o xmo.F .1411. .1 11 11111111111, 111111(\1 mcwpmmmm o o P- v- _ open 40 > p m 11 H u m z z < m h e u z 3 H o z 4 o . u o 1111. . immolimizee - 1 ll >o.n Fucenopxmp. 207 .eoeoaepeam .momn we; muumeem cowaapom :og on one mucmuuzucou ucaocmxoam .Focegum c? :ox zepmmo.o new: omza cw pageane zommo.o mo :oeuumma on» com mmccno mucmuuaucou .eo oc=m_a Rummv meek . mN o o .c -opxm.m a11 a a a q _ .4 a a _ F m IL UOZQSUPgzo covuumpmm :uvwz mmpam $0 copmgm> mgmzmo .no mgsmwd Mgr ME. ME (i , a“. E _., rLJ\ 223 graphics editor contains provisions for searching the buffer for a given X or Y and outputting any point number where this value is encountered. In addition, editing options for changing points and scaling factors, and inserting or deleting points, are provided. Finally, the graphics files created with FILEII may be test plotted by FILEII option 3. All of the text and graphics files used in CBHELP were generated with FILEII. There are currently 59 text messages and lO figures used with CBHELP which were produced and incorporated into CBHELP itself with only the addition of three lines of programming to CBHELP (and, of course, recompilation). Turn-around time in the use of CBHELP: Because file-structured data sets were used with CBHELP, and because those files are accessed via DECTAPE, the time involved between the selection of an option and its actual appearance to the user can be as long as 20-25 seconds. This lack of high-speed acquisition of files is probably the singular non-pleasing aspect of CBHELP. Faster acquisition could have been achieved with the use of RTAPE, which does not need to consult the directory before reading the tape, but, as mentioned above, this speed would have been achieved at the expense of flexibility and expandibility of the program. This problem will be completely overcome if the program is ever transferred to a disc-based system. Core requirements: Both FILEII and CBHELP require 12 K of memory for operation. 224 C. USING THE CBHELP PROGRAM It is not practical to present a flowchart of the operation of the CBHELP program in this manuscript because of the very large number of cross-interactions and branches available within it. A listing of the CBHELP program itself appears in the Appendix, however, the text messages and graphic displays, because of their large number, have not been included. CBHELP can probably best be illustrated by guiding the reader partially through the program so that he might develop an insight into its operation. The canister in which the CBHELP tape is contained holds instruc- tions which should guide even the novice operator through loading the tape and calling the program. When CBHELP first responds, it asks the operator if he would like a list of instructions to follow in using the program. These instructions include turning on the scope, plotter, and line printer, how to type answers to questions, choosing an option within CBHELP, etc. If the operator wishes to view such instructions, he types "YES" and they are given to him. CBHELP follows with an initial dialog, that can be listed on the teletype, line printer, or display scope, or skipped if the operator is already familiar with the program. Next, the major options available with CBHELP are printed and the operator is asked to select one. This dialog is shown below: PLEASE TYPE THE NUMBER OF THE OPTION YOU WISH TO SELECT: 1) GENERAL SYSTEM DESCRIPTION 2) DESCRIPTION OF THE INSTRUMENT 3) DISCUSSION OF CONDUCTANCE MEASUREMENT 225 DISCUSSION OF THE BIPOLAR PULSE TECHNIQUE INSTRUCTIONS FOR ROUTINE OPERATION INSTRUCTIONS FOR PULSE WIDTH SELECTION INSTRUCTIONS FOR ROUTINE ADJUSTMENT ERROR MESSAGE SUMMARY AND TROUBLESHOOTING SELECTION OF A PROGRAM SET FOR SPECIFIC EXPERIMENTS CREATING OR CHANGING A CBHELP FILE TERMINATION OF THIS PROGRAM —l—l HSOQNGm-fi WWW Ito When the operator types a particular option number, the program branches to that subject and calls the proper files from the knowledge banks on tape for graphics displays and text messages. For example, if the operator selected option 9, because he wishes to utilize the computerized conductance system and its program set to perform a specific experiment, the computer would respond with: 9 THE NAMES OF THE PROGRAMS IN THE COMPUTERIZED CONDUCTANCE SYSTEM ARE CODED FOR EASY RECOGNITION. WOULD YOU LIKE AN EXPLANATION OF THIS CODE? YES OUTPUT ON LPT=l, ON TTY=O:1_ If the operator types "YES“ an explanation of the program nomenclature is generated. If he types "NO", the program moves immediately to output the list of available program sets, shown in Figure 68. If the operator were to select, for example, program set 3, CBHELP produces the tem- perature-conductance program flow chart shown in FIGURE 69. The operator is then asked: WOULD YOU LIKE AN EXPLANATION OF THE PROGRAM SET? If the operator answers "YES" such an explanation is produced for him. It is in this way that multi-level explanations are built into CBHELP. 226 I)A TITRATION 2)A DETERMINATION OF DISSOCIATION CONSTANTS: EQUIVALENT IONIC CONDUCTIVITIES: OR THE RELATIVE STRENGTH OF AN ELECTROLYTE 3)A TEMPERATURE-CONDUCTANCE PROFILE FOR DETERMINATION OF CORRECTION PARAMETERS TO BE USED WITH VARIOUS DATA ANALYSIS ROUTINES: OR FOR A QUALITATIVE OR QUANTITATIVE EXAMINATION OF THE CONDUCTANCE BEHAVIOR OF AN ELECTRO- LYTE UITH CHANGES IN TEMPERATURE. THIS SEQUENCE IN- CLUDES THE ARRAY ARRANGER ROUTINE FOR LINEARIZING A NON-LINEAR TEMPERATURE CHANGE TO PREVENT VEIGHTING OF DATA IN ANY PARTICULAR TEMPERATURE REGION. 4)A TEMPERATURE-CONDUCTANCE RUN AS IN (3) BUT UITHOUT ARRAY ARRANGING THE DATA (USED VHERE THE TEMPERATURE CHANGE HAS BEEN LINEAR). 5)A KINETICS RUN IN WHICH ONLY CONDUCTANCE AND TEMPERATURE ARE TO BE MONITORED AND IN WHICH ONLY CONDUCTANCE IS TO BE PLOTTED. 6)A KINETICS RUN IN WHICH AUXILLARY DATA: IN ADDITION TO CONDUCTANCE AND TEMPERATURE DATA: IS TO BE TAKEN: OR IN WHICH TEMPERATURE DATA WILL BE PLOTTED. 7)CHROMATOGRAPHIC MONITORING 8)INSTRUMENTAL NOISE DETERMINATION 9)INSTRUMENTAL ACCURACY DETERMINATION IC)AVERAGE OF N DISCRETE MEASUREMENTS II)INSTRUMENTAL FUNCTION TEST Figure 68. Program Set Selection Options in the CBHELP Program. 227 THIS IS THE PROGRAM FLOW CHART FOR THE TEMPERATURE-CONDUCTANCE DATA ACQUISITION AND ANALYSIS: WHICH USES THE ARRAY ARRANGER: CBTALR: TO PREVENT ACCIDENTAL UEIGHTING 0F POINTS DURING THE CURVE FTTING. *¥#***#*.¥ * CBTSLS * *t**#**#$* tfiitfi$fitii 41 CBTALR t **¥*$***$. coo *t***¥*$¥. :11 CCLALF t ttt¥¥¥¥l¥¥ co. 000 000 000 .00 o o o o o $*$.#*$*.¥ Itfitfififi‘.. #fifitttttlt itttfiitfifit ttt¥$$.¥*‘ * CDTALL * * CDTALC i * CDTALC * * CDTALF * t CDTALS # **#*#’¥’.. ....*’*.¥* ****¥*¥*$* ¥********* ****#*#$‘* 0 o o o o ooooooooooooooooooooooooooooooooooooooooo o 000 o 0 *¥.#.‘*... 0 o t CFPTLI I o o .¥*$**..*. on...ooooooooooooooooooooooo o o one no. a o .$*.*O.*.t fit.¥.¥*.‘. * CELTLR t t CFPTLR t ¥.*¥***‘*‘ .##*.#.... Figure 69. CBHELP-Generated Program Flowchart for Temperature- Conductance Profile Measurement and Analysis. 228 If the operator types "NO", he is able to select another of the ll programming options or he may return to the selection of a different one of the major options. If he were to choose other options within the option 9 program selection routine, he might receive flowcharts and explanations for other types of experiments, and even graphic displays depicting a sample titration curve with explanation of its various regions, or a sample strong-weak electrolyte comparison curve and so forth. Any of the other major ll options produce figures and text and interactive dialog as does option 9. CBHELP will even check to see if the line printer is turned on and tell the operator to activate it if he has requested output on it and has failed to turn it on previously. It can be seen that most of the subjects which one would be interested in with respect to utilization of the computerized conduc- tance system are included in CBHELP. Admittedly, CBHELP lacks much of the sophistication of many institutional computer-based teaching systems such as, for example, the Plato IV system at the University of Illinois (67). CBHELP does go much further in using the computer system to instruct an operator in the use of the instrumentation con- trolled and monitored by that computer system than any device which computer-based instrument manufacturers have provided with their systems. It serves an important function by largely assembling the theory and techniques of the computerized conductance system into one package to present to the prospective or veteran user. In addition to this major purpose, it is possible that CBHELP will point the way toward instrumentation systems which are still more interactive with the operator. There is much to be gained in continuity of instrument use, 229 a steady source of reference material, and user acclimation to a new system by permitting "smart" instrumentation systems to tell their own story to prospective "clients". It seems somewhat short-sighted to build an elaborate and useful computer-interactive laboratory. measurement system and provide with it stacks of manuals in conventional form. Such systems are often so complex that it is impossible for a casual user to delve through these mounds of literature, or even for a dedicated user to troubleshoot or expand his use of the system efficiently. Thus, it is hoped that CBHELP will be viewed as movement toward more complete user-system interaction, where the system is capable of that level of interaction. CONCLUSION The capabilities of the computerized conductance system have been demonstrated in a significant number of diverse applications. It has proven to be a sufficiently powerful system to arouse the interest of other workers who will benefit from its continued usefulness as a dynamic measurement system. Because of the nature of the system itself, it is impossible to state that either the instrumental work is complete, or that all areas of possible application have been defined. Although the hardware portion of the system has remained relatively unchanged during these studies, it is possible that some modification of the circuitry will be undertaken by later workers to make it operate with other computer systems, expand its operating range, possibly incorporate bipolar current techniques for high conductance measure- ments and so forth. The author has shown that the system, as it developed over the past three and one-half years, included a number of innovations which give it the flexibility to perform meaningful measurements for operators of various backgrounds and interests. Finally, the success of the project serves to legitimatize the pursuit of instrumentation research for chemists by showing that such research can produce devices which not only perform sophisticated measurements for the scientist with an instrumentation orientation, but continue to be of use to other workers as important laboratory tools after the instrumentation chemist has completed his work with that particular measurement system. 230 REFERENCES 0101-wa \l 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. REFERENCES Controller design by Hahn, B. K., Interface design by Kelly, T. Interface design by Rabb, M., and Hahn, B. K. Interface design by Davis, R. and Hahn, B. K. Designed by Last, T. ' Hahn, B. K. and Enke, C. 6., Anal. Chem. 36, 651A (1973). Heath-Schlumberger product bulletin 595-1422. Computer Interface A09. (1972) Program written by Hahn, B. K. Livingston, J., Morgan, R., and Lammert, O. M., 9, Am, Chem. S99, QB, 1220 (1926). Kohlrausch, F.,Wied. Ann, 62, 249 (1893). Johnson, D. E. and Enke, C. 6., Anal. Chem. 52, 329 (1970). Jones, G. and Bollinger, G., J. M. Ch_e_m_. Soc. 613, 411 (1931). Washburn, E., and Bell, J., J, Am, Chgm. Soc, 33, 177 (1913). Washburn, E., J. Am. M. £13. 3%, 2431 (1916). Jones, G.and Josephs, R., J. Am_. gm. S_oc_. 1119’ 1049 (1948). Shedlovsky, 1., g_. m. C_hgm. S_og. 3g, 1793 (1930). Schmidt, K., Rev. Sci. Instr. $1, 671 (1966). Bentz, A. J., Sandifer, O. R., and Buck, R. P., Anal. Chem. 36, 543 (1974). Johnson, Donald E., Ph.D. Thesis, Michigan State University (1970). Daum, P. H. and Nelson, D. F., Anal. Chem. 53, 463 (1973). Philbrick Researches, Inc. Applications Manual for Computing Amplifiers, Boston: Nimrod Press Inc., 1966. 1.26. Perone, S. P., Anal. Chem. éé» 1288 (1971). Frazer, J. W., Anal. Chem.‘%g, 8, 27A (1968). Venkataraghavan, R., Klimowski, R. J., and McLafferty, F. W., Accounts Chem. Res. 3, 15B (1970). 231 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 4o. 41. 42. 43. 44. 45. 232 Dessy, R. E. and Titus, J. A., Anal. Chem. 45, 2, 124A (1973). Dessy, R. E. and Titus, J. A., Anal. Chem. 46, 3, 294A (1974). Ramaley, L. and Wi1son, G. 5., Anal. Chem. fig, 606 (1970). Parker, R. A. and Pardue, H. L., Anal. Chem. 44, 1622 (1972). DeVoe, J. R., Shideler, R. W., Ruegg, F. C., Aronson, J. P., and Shoenfeld, P. 5., Anal. Chem.lgg, 509 (1974). Heath-Schlumberger equipment bulletin 595-1422, Computer Inter- face ADD , "Modular Component Cards" (1972). Lauer, G., and Osteryoung, R. A., Anal. Chem. 4Q, 10, 30A (1968). Perone, S. P., Jones, D. O., and Gutknecht, W. F., Anal. Chem. 4A, 1154 (1969). Keller, H. E. and Osteryoung, R. A., Anal. Chem. 43, 342 (1971). Perone, S. P. and Eagleston, J. F., 9, Chem. Ed, 48, 317 (1971). Toren, E. C.,Jr., Carey, R. N., Sherry, A. E., and Davis. J. E. Anal. Chem.lgg, 339 (1972) Hicks, G. P., Eggert, A. A., and Toren, E. C., Jr., Anal. Chem. 42, 729 (1970). Keller, H. E., Courtois, G. E., and Keller, J. E., Chem. Instr. 4, 269 (1972). Digital Equipment Corporation. Introduction 59 Pro rammin , Maynard, Mass.: Digital Equipment Corporation, 1972, Ch. 9. Digital Equipment Corporation. Programming Languages. Maynard, Mass.: Digital Equipment Corporation, 1972. pp. 13-37 to 13-39. Ibid. pp. l3-39 to 13-42. Ibid.p. 15-80. Ibid. pp. 15-73 to 15-76. Ibid. p. 15-80. Kelly, P. C. and Horlick, Gary, Anal. Chem. 45, 518 (1973). Malmstadt, H. V., Enke, C. G., Crouch, S. R., and Horlick, G. Optimization of Electronic Measurements, Module 4, Menlo Park, Ca1if., W.TA._Benjamin, Inc., 1974. Digital Equipment Corporation. "MAINDEC Test", 1968. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 233 Hahn, B. K., unpublished program. 1972. Hildebrand, F. 8., Numerical Analysis, New York: McGraw-Hill Book Company (1974). pp. 549-559. Franklin, E. C., 9, Phys. Chem.‘L§, 675 (1911) Armitage, P. T. and French, C. M., A, Chem. Sgg,, 743 (1963) Falkenhagen, H., Electrolytes, London: Oxford University Press, (1934), pp. 200-201. Hall, J. L., Gibson, J. A., Jr., Wilkinson, P. R., and Phillips, H. 0., Anal. Chem. 26, 1484 (1954). Farrow, R. N. P. and Hill, A. G., Analyst. 23, 210 (1965). Levine, S. L. and Golden, H. J., Anal. Letters, 1, 39 (1967) Bauer, W. E., private communication. Albrecht, H. 0., Z, Rhys. Chem. L36, 321 (1928). White, E. H. Zafiriou, O., Kagi, H. H., and Hill, J. M., 9, Am, Chem. S39, 86, 940 (1964). White,E. H. and Bursey, M. M., Ibid., p. 941. McCapra, F.,Quarterlngeviews, 485 (1966). Beck, M. P. and Joo', F., Photochemistry and Photobiology,‘LQ 491 (1972). Gorauch, J. D. and Hercules, D. M..Photochemistry and Photobiology, L5, 567 (1972). Lee, J. and Seliger, H. H..Photochemistry and Photobiology,,1§. 227 (1972). Drew, H. D. K. and Garwood, R. F..g, Chem. §gg,, 791 (1938). Seitz, W. R. and Neary, M. P..Anal. Chem. $6, 188A (1974). Digital Equipment Corporation, OS/8 Software Support Manual, Maynard, Mass.: Digital Equipment Corporation, 1972. p. A-5. Digital Equipment Corporation, Programming Languages, mp, git. pp. 13-18 to 13-20. Hahn, B. K., XYSYS.SB and SCOPE.SB Plotting Routines, unpublished programs. Smith, S. G. and Ghesquiere, J. R. Computer-Assisted Instruction in Chemistry, Part B: Applications, ed. Mattson, J. 5., Mark, H. B., Jr., and MacDonald, H. C., Jr. New York: Marcel Dekker, Inc., 1974, Chapter 2. CBTSLS PROGRAM LISTING [PROGRAN NANEI CBTSLS.FT [FORTRAN-SABR! KEITH J. CASERTA 1974 ’fugg [5 14: 3:“EQAL FORM 0: THE BASIC DATA AQJISITION PROGRAM IFOR 74E OONPJTE411E -J CJVDJCTANCE SYSTEM. IIT aes1vs 41T4 A 9%: LInINAav SCAN OF THE CELL. DURING WHICH [THE O’TIMIZED CIRCJIT PARAMETERS rOR EACN OF FOUR PULSE IHTDTRS ARE DETERMINED AND STORED. [THESE VALJES ARE PRINTED. THE AVAILABLE OPTIONS AT THIS POINTI [1)AVERAGE OF AN [NPJT vUNaEi OF SCANS /2IRESTAeT ISITIHED DATA AcauSITION OF UP TO 500 aOINTS OF UP TO I 2047 ACJJISTTIJNS PER ’OINT FOR 30TH cOND AND TEMP I I’ ZERO IS TNPJT AS THE 4 OF TEMP POINTS. TEMP MEASUREMENT I Is S0077 [AC(0017. GO FIX [IS A/D NEAR FULL ScALE. [T559 30 FIX INSTRUCTION MAY BE INSERTEO INTO 0A1? "ao ALL DATA POINTS VET: IND: CONTINUE IVES: OUTPUT [SKIP 3N CLOCK OVERFLOH [SRIP ON ONERFLDN ERROR [NO ERROR. RETURN FOR NEXT POINT [OO INDICATE CLOCKINS ERROR [.OC 07 FLYING ADD LSD [-Oc DE FLYING ADO MSB [NUMBER OF ACOUSITIDNS POINTER [SECONES RESETTABLE C ACOJISITIONS POINTER [NUNaER of DATA ’OINTS POINTER [OUTPUT ROUTINE ADDRESS [SURRENT MEASURENENT TD 3E CHECKED FOR OVER/UNDER RANGE [ADDRESS OF INCREASED RCELL CONPENSATOR ROUTINE [ADDRESS OF DECREASED RCE.L CONPENSATOR ROUTINE [CONTAINS I OF TENP POINTS POINTER [CLOCAIND ERROR INDICATINO ROUTINE LOC 245 I [THIS ROUTINE RESETS THE STSTEM TO COMPENSATE FOR INCREASED RCELL PAGE EAP O'II1:3LA TAO TAD SNA JNP 0'2: TAD DCA 3LA TAO JNP OF12: OLA TAO TAD SZA JNP CLA TAD DCA JNP GOA: 35A TAO RAL SZL JMP DCA IAC JNP 0'14: OLA TAD SAL DCA TAO JNP CLL JNITS CF1000 CF12 C7400 JNITS CLL 37760 3A2! CLL IVP (7570 COA CLL (4010 IVP OPZ CLL 3NP 3714 OUR OA20 CLL IVP IVP 30004 OA20 [NININJM OFFSET. [vEs. OD RESET PR [ND.'SET 1 LESS UNIT [OUT 1 OFFSET UNIT VALUE [:NECN FDR NAx GAIN SITUATION [RETURNS 0 Ac AT MAX CAIN [NOT AT NAx OAIN. CONTINUE [NEN'I/V Ir MAX GAIN [RENOvE LAST OFFSET UNIT. SET PARA [INcREASE PR IF POSSIBLE [3°NE T00 EAR: IVES. CRECK CAIN INO: pd SETO STORE [A001 [INCREASE GAIN IF POSSIBLE [INCREASE GAIN VALUE IN PARA HORD I [THIS ROUTINE RESETS THE SYSTEM TO COMPENSATE FOR DECREASED RCELL I OFIx2:3LA TAO TAD SZA JNP CLA TAD DCA 008: 3.A TAD TAD SNA JNP TAD DCA CLA TAD JNP OF21: 3-A CLL IVP (3770 COD CLL (0210 IVP CLL JNITS 15400 OPEI OZOOO JNITS :LL OOOZD OA21 :LL [CNECK FOR PREVIOUS 0 OFFSET [RETURNS 0 AC AT PREVIOUS 0 OFFSET [NOT AT D. CONTINUE [NEN'I/V SETTING [NAxINJN OFFSET. IVES. 30 RESET GAIN ’Nao SET 1 "ORE UNIT [JNITS INCREMENTER FOR PARA NORD uuwmumumu mumwmmuwmuuuummmuumuauuu {DID «Snowman muaamuuaauauaonauu 0'22: [BEGIN RESET ROUTINES. TAO RAR SZL JNP DCA TAD JNP CLA TAD RAR DCA TAD TAD SNA JNS 5(P 4LT OLA CMA JNP IVP 3F22 IVP 37774 3A22 CLL aNP OHP Sup 37400 I 0'19 CLL 3A22 246 [DECREASE GAIN IF ROSSIDLE [GONE TO FAR. IVES. 30 LOHER PM IF POSSIBLE INO. STORE [DECREASE PM IF °OSSIOLE IOONE TOO FAR. [30 TO SUBSCALE RARNINO ROUTINE [TESo STOP [NO. SET UP [-1 EXECUTED AFTER RETURN FROM OFIX ROUTINE [ROUTINE T3 RESET AFTER aETJRN FROM DFIX1 FROM PM. GAIN CHANGES OAZO. [THIS OA21: [TRIS OA22: [THIS 0A23: 0A24: OA25: TAD AND TAO DCA TAD c. JMP SPARA 32MS( 30240 SPARA 32400 JNITS 3A2! [RESET PARA HORD oN RETURN FRoN FIX [REsET OFFSET UNITS IN PARA RORD [SET OFFSET UNITS I 10 IN PARA NORD [SET UNITS POINTER - 10 OFFSET UNITS [3O LATCH CIRCUIT Is RETURN ROUTINE FOR RESET OF UNITS ONLY TAD DCA JNP SDARA SPARA 3A24 [RESET PARA HORD [3O LATCH UNITS IS T45 RETURN ROUTINE FOR OFIXZ FROM PH, GAIN CHANGES TAD AND TAD DCA TAD DCA SPARA 32H3< -0020 SPARA 27000 JNITS [RESET PARA UORD [REsET OFFSET UNITS IN PARA NDRD [SET OFFSET UNITS O 1 IN PARA 4000 [SET OFFSET UNITS POINTER s 1 OFFSET UNIT ROUTINE RESETS CONRBIP FOR NEH PARAMETERS OLA TAD TAO .PHH CLA TAD TAO C.A JNPI 07000:7"00 02400:ZROO 00240:.240 OZHSK:7417 LOO2D. .220 02000.2000 07700:7760 00004:;004 07400.7400 CLL 34p ’dP CLL IUP JNITS CLL 3F16 [RESET PH. PH [.ATCH PH. RH [RESET UNITS. OFFSET DECADE. GAIN [-ATCM [RETURN TO MAIN aROO [OIVES OFFSET UNITS - 1 [10 OFFsET UNITS VALUE TO CONPBIP [PARA dORD 1n OFFSET UNITS INDICATOR [PARA JORO OFFSET UNITS VALuE NASK [1 OFFSET UNIT VALUE TO CONPSI’ [JNITS INCREMENTER [PARA dORO OFFSET UNITS DECREMENTER [PARA dDRD GAIN INCREMENTER [8N UNDERRANOE CNECK 247 O H5A00.3400 [OFFSET OVERRANGE CHECK 3 00020.4020 [3ARA JORO OFFSET INCREnENTER 3 07774.777. [3ARA aono GAIN OECRENENTER SOF1000.1"00 [NIN OFFSET I'II CHECK S 071‘. 3A17 : 0F19. FALTz [SUBSCALE NAININO ROUTINE [ S [SUOPROORAN DTOUT 3 [THIS SECTION SETS U’TTNE TRANSFER To THE ANALYSIS ROUTINE 8 [ 3 OTOUTozLA SLL NARAIZOASIINT NARAIZOAAIIITRVL NARAIZOASIILD NARAIZOASIII’TS CALL uTAPEIoniL(.2046.NARAI :ALL EXIT I [SUBPROSRAN FA;T1 [THIS SUBPRDGRAH INDICATES THAT CELL OONO IS VERY LON [ FALTlazLA SLL ARITEI1.1leN 4 FoNNATI[.vKROa 1 AT =4 - '.F6.3.' HSEC'I OLA :LL JNP I PAF1 PRFI. AN [ [SUBROJTINE FALTZ [THIS SJRROUTINE INDICATES CELL COND TOO HIGH [ FRLTZDJDOO OLA :LL dRITEI1.15IPH . ‘ 5 FDRHAT([.IERR3R 2 AT Pd . '.F6.3.' NSECII OLA SLL ISZ ’ALT2 [SET TO RETURN TO MAIN FROG ¢ 1 '17. JND I F‘LTZ I [SUBPRJSRAN FR;T7 [THIS SJBPROGRAM INDICATES THAT ONE OVERFLOH DF CLOCR TIMING [HAS OZCURRED DURIN3 TDA - A FATAL ERROR HHICH RALTS THE PROGRAM F‘LT’Oag‘ :LL dRITEI1.117I 17 FORNATI[.o§qR3a 7') 4LT I [SUBROJTINE SETU° [THIS SJBROJTINE SETS NECESSARY PARAMETERS FOR AVERAGING [A SERIES 3' SCANS! AND SETS PROPER PH. AND INSTRUNENT PARAMETERS OGOWUIIDUDH mum manna"... “ADOOOCUUUP unaut- [ SETUPAJOOO SEA0I1.TIA°N.APTS.LO 7 FORNATI'AT RHAT Pd. INSECII '.EIO.B.[.'NJNBER OF G SCANS TO AVER OAGEI '.E16.0.[.'JDU3LE aRECISION- I1'T.O'NII'.I5I JRITE(1.309) . 309 FORMATI'TVPE ESPEQINENTAL INFO ISNTRL 0 TO ENOII'.[I 3 TEI1. :LA SLL S TEIZ. (SF [NEYSOARO STRUCK YET. Ena'uun‘.“HMEflUOfl5 GD“,“HDIDUI“HDGDUDUOMHDIDUOUH.(DUIMHDRDUIGDMD (DID “ND SQUI I. TEI3. TEIQ. 1O 11 H7. H0. H“. JNP (RB TL! TAO SZA JNP JNP 7571 OLA OLL J°U1 lx-.OO1 XXIXKOlfl. TEI2 TEIJ TEII TEIQ 248 [NO CHECN AGAIN [vEs. READ CHARACTER [ECHO IT ON PRINTER [-207 [IS IT IN ALT MODE. INO. RETURN FOR NORE IVES. CONTINUE IN NEXT ROUTINE [-207 IFIA’H'XII11.11.10 JPIJ3-1 30 TO 9 IRTSIIFIXIAPTSI OLA OL NluslPH'IJE) TAO ININ OOA aRP NIIIIIV'IJEI TAD INII OOA IVP NIHIIPH’IJPI TAO ININ DOA 3UP VIUIIUNITIJPI TAO INIJ DOA JNITS TAO IVP TAD JNITS -IvO ISZ JNP ISZ JNP OLA TAO TAO 59"" IS! JNP ISZ ASET JHP 4R O-A O! JNP I‘SETUF ASET .E ASET 4r OLL ’RP ’HP ASET d3 I [SUBROJTINE TYPO [THIS SJRROJTINE READS TTv AND CHECKS FOR A ‘O' I TYPGC AR. AS. TKO VTATI SZA JNP AA JNp I Tvoa V7471.7471 [.OAO PR /-OAO I/v. OFFSET DECADE [-OAO Pu I-OAD OFFSET UNITS [-ATCH OFFSET. Ilv [.A;T FOR RELAYS To SETTLE /-RTCH PH. PH [INIT FOR RELAYS TO SETTLE [RETURN TO HAIN FROG I YES '3' TO START'I [(EYSOARD STRUCK YET. INO. CNECK AGAIN IVES. READ CHARASTER [ACKNONLEDGE IT ON PRINTER [SJBTRACT 307 [IS IT A.”O'. IND. 33 CHECK AGAIN IVES. RETURN TO NAIN PRosiAn [-307 (3\‘\‘-\ CODE. 061. NSKI. 062. 063. HSKE. SDC777A. 3 HSKS. 3067760. S DOG. O0 O11 O12 [ [SUBROJTINE OOOOE [THIS SJBROJTINE DECODES PARANETER HORDS AND cONPUTES [PH.RV.XUNI.ES.RCELL.ANJ COE.L. IT NUST BE ENTERED HITH PARA HORD IN Ac [PARA dJRD FORNAT. JOOO OOA SPPJ NONI-O. ’NI.OOS Rv-1ODOO. TAD SPPJ AND NSKI' DOA SPPd TAO SPPN SNA JNP 0C2 OLA OLL RHIP4010. ONA TAO SPPN JNP OO1 3003 TAO SPPJ AND NSNZ OA SPPd TAD 399d SNA JNP DCA OLA OLL RVIRVI10. TAD 0:777. TAD SPPd JNP DDS 3314 777. TAD SPPJ AND NSKS DOA 399d TAO SPPd SNA JNP OCO OLA OLL ‘UNIOIUNIOI. TAD 987750 TAD SDPN JNP DDS 3360 7760 3;A OLL 249 FIRST 4 BITS (H53) ARE UNUSED NEIT A BITS FOR JNITS VALUE (BINARY 1'10) NEST 2 BITS FOR SAIN INFO (BINARY VALUE 003’ EAT 2 BITS FOR RH INFO (BINARY 1-3) [STORE PARA HORD [NASK ALL BUT PH BITS [ALLSET. [vEs, OONTINUE IND. ADJUST [ADD -1 [SET UR NEXT ITERATION [NASN ALL BUT PH BITS [RASN ALL BUT GAIN BITS [ALL'sET. /Y§S. :ONTINUE /N°.'A°JUST [SET U, NEXT ITERATION [NAS( ’OR ALL BUT GAIN BITS [SAIN OEINCREHENTER [NASA ALL BUT OFFSET UNITS BITS [ALL SET. IVES. CONTINUE IND. ADJUST [SET JR FOR NEXT ITERATION [NAS( FOR ALL BUT UNITS BITS IJNITS DEINCREHENTER ESIIZOI12.5[‘096.I°6.25) IFILO)312.312.511 ESL'IZL'I12.5I4095.IOO.25IIAPTS ESIESoES. RCEL.IP4II(ES[RVIOI(10.0XUNIIIRVI) :CEL-I1./QSELL OLA SOL JNP I OODOE END CBPSLT PROGRAM LISTING “HQUNCUICHDUOUH‘UOCH.IIUHD“MNIICHDIIUHDIO‘HIINIOI ‘\‘\\H\‘b\‘§\V\‘5\H\‘h\fl\‘\\r\‘\\H\‘§\.\J§‘s\ [PROGRAN NANE. [FORTRAN-SARR. NEITN J. CASERTA OS'SLT.FT ‘IRITA [TRIS ’RDGRAN RUNS dITN A~V ONE OF FOUR DPTIONSI IISEARCN A RREVIOJS FILE GENERATED HITH THIS PROGRAH AICALCJLATE NAXINUN AND NININuN ERROR SICALOJLATE AVERAGE ERROR OIDuTPJT THE DATA SET ON THE LPT ZIPLOT A PR5VIOUS-Y SEVERATED FILE AICALOJLATE AND OUTPUT HAXINUN ERROR SIREDUEST SCOPE SCALING aARA-IETERS OIPLDT POINTS dITHIN LIMITS 3ICDN'ARISON RJN AIREDUEST NuNSER OF AVERAGES DI'ERFORN ’RE-ININARv SCAN. OUTPJTTING A *.~ HHEN DONE :IQEDJEST REA- R FOR I AXIS (AS .OGIRII DIDPERATOR SE-ECTS OPTION: NEASURE REASURE AND AVERAGE INRuT SCANS OF THE STANDARD AND AVERAGE INPUT SCANS OF THE COMPARISON RESET X AXIS VALUE :0 ON TO NEXT VALUE. STORING CJRRENT DATA :HAVGE 34 END NEASJREHENT AND SELECT O’TIDN OUTRUT'ERROR RIVARIASLE INORENENT RJLSING AI’ERFJRN ’RELININARN SCAN. OUTPUTTINO A ".I HHEN DONE SIREOUEST INTERVAL. USED AS X AXIS (AS LOGIINTERVALI) OIDPERATOR s AOOJIRED AS 1000 SCANS AT THE CHOSEN INTERVAL ODNNON OARv,NARA DIMENSION DARVI2.501.NARA(2I OPDEF OPDE’ OPDEF OPDEF OPDE’ SNPD’ OPDEF OPDEF SNPO’ OPDE' OPDEF JPOE’ OPDE’ OPDEF flflfiE? OPDEF OPDEr OPDEF 330E? OPDE: OPOEr June: OPDEF OPDE’ SRRDT SNPO’ OPDEF S(PD’ SRPD’ OPOE' OCFP. ODOR OCFDD ECPI OCFOD TCFL TDST TRIG TTFL LIVD -PHH ET°I OTFDG TTAo TFPGT TIT -PseT OLCL ICNTR OLCIC LCNTR RCTRL RCNTR LCTRL SKPDF SKPDE OLOFE SKPTO SKTgE OLTDE 6321 6322 6323 6324 6527 6331 6332 633A 6541 6342 634. 6351 6352 6354 6356 6361 6121 6122 6123 612. 5125 6126 6127 6131 6132 6133 6134 6135 6136 6137 cTS OPTION As IN (D) ABOVE BUT DATA IS THIS aROGRAN IS BASICA.LV DESIGNED FOR ACCURACY DETERMINATIONS [CLEAR CONO FLAG. CLEAR PI ENABLE [GATE CONDUCTANCE DRIVER [CCFP - OCOR [ENABLE OOND PROGRAN INTERRuPT [CCFP - OCOR . 53’! [TEST CONDUCTANCE FLAG [TURN OFF SEOUENOE TRIGGER [TRIGGER SEQUENCE [TEST TENPERATURE FLAG [LATCH IIV. OFFSET [LATCH PR. PH [ENABLE TENP RRoORAN INTERRUPT [CLEAR TEMP FLAG. aI ENABLE. GATE DRIVER [TRIGGER TEMP A/O CONVERSION ICTFPG O TTAO [TITRATE [LOAD THE CLOcN PRESET REGISTER [CLEAR TNE CLOCK [INITIALIZE THE COUNTER [CLEAR C.OCK AND INITIALIZE COUNTER [LATCH TNE COUNTER [READ THE COUNTER LATCN [READ THE COUNTER [LOAD CONTROL REGISTER [SKIP ON DVERFLDJ [SKIP ON OVERFLDR ERROR [CLEAR OVERFLoN AND OVERFLOH ERROR FLAGS [SKIP ON TIME BASE FLAG [SKIP ON TIME BASE ERROR FLAG [CLEAR T.a. ANO T.B. ERROR FLAGS 250 OPOE’ sans, 0305' OPOE’ sans: ans= Oooer aaoe= OPOE' ans: OPDE’ acne: ASSVN Aaqu AESYN Aasvn 035'! 038YN ”UV DVI NNI SHL SSH LSR NOL SCL SCA NOA £3085 STORE a”: In. TV“ JNITS ASET SPARS 7409 7407 7411 7413 7415 7417 7421 7403 7441 7501 6034 6057 0074 0075 0076 0077 0100 0101 25] [NULTIPLV [DIVIDE [NDRHALIZE [SHIFT LEFT [ARITHHETIC SHIFT RIGHT [LOGICAL SHIFT RIGHT [LOAD HU.TIPLIER OUOTIENT [STEP COJNTER LOAD FRON HEHORV [STEP COJNTER LOAD INTO AC [HO LOAD INTO AC [ERASE DISPLAY SCOPE [SET SCOPE IN STORE NODE [PH POINTER [PH POINTER [IIV POINTER [OFFSET JNITS pOINTER [GENERAL POINTER [PARAMETER POINTER [INDIRECT ADDRESS IN FIELD 1 POINTER [INDIRECT ADDRESS IN FIELD 1 VOINTER REST! SPPU 0102 RESYN SP’H 0103 :muuuuuauauuauaumauuuouuau [ I [PROGRAN BEGINS NTTR FIRST D’TION SELECTION [ I NTIO 172 4R1TE¢1.372)RT. 072 ’OPHATII.'SEAR:H aREVIOUS FILE I 2'.I.'PLOT PREVIOUS FILE . 1'.[ ’A'START CONRARISON c O'.I.'VARIA8LE INCREHENT °ULSING - -18'-I01 aEADIIAISIIJO 161 'ORHATIISI IPIJOI1SDA160.159 159 REA011.1501J 153 FORHATI°FIR5T 3L03410.151 3ALL RTAaE¢1.J.374.OAaVT IFINARAIZII3OOA3310301 360 dRITEI1oSPZ) 382 =ORNATI'THIS IS A OONPARISDN FILE'./) 30 TO 394 301 dRITEI1A3O3) 363 ’ORNITI'THIS IS A VARIABLE PULSING FILE'./I 304 IINARAI1I IFIJO-21519.502.532 619 JOINARAIZI S OLA :LL S JNS AOUT S JNS TOUT S OLA CLL 30 TO 69 :60 OONTINUE [ 3 [BEGIN THE ERE;IN1NA?Y SCAN SEOUENCE S I 3 PNNVO :5‘ :LL ' S DCA PUP [SET Pd AT 0.01 NSEC SEAOIIA7IPN 7 'ORNATI'EH INSECII'.EIz.OI IFIJ51967.918.366 966 READII.3691S1S.319.519 S13 INTN-OAav(2.J) TNTN-OANN(1.JT S10 :ONTTNUE ANTT§c1.S14TXNIx-V*Afio'H!"-'"'" S14 '0RHAT(I.TNAX in I 'oE16.0.' AT x I '.E16.0./.'HTN ER I '.E16.S. 0' AT x I '.E16.S) S20 NEAO¢1.521TJ S21 ’ORHAT('DRTTOVSItoRESTARTIDT'AIST 3 SFCJ’bSASSASOZ I : ITH!S IOUTTNE 'EQFOQSS THE AVERAGE OF AN INPUT NUMBER or SCANS I S AVERGOO100 S ‘00 3LA SLL STI-1 0 :LA :LL _ S TAD I<0 [SET UP DIVTSOR S :A VSAR S TAO I(D ISET 0’ a or POINTS POINTER S :TA - . S 03A 39A! S 03A N-SS ISLEAR LSB NORO S :A NNsa :LEAI N58 NORO S A“. 31A SLL S Talc I?ULSE TRTGGER S Ax. TSFL [TEST 'LAG S JNI A! S SCPPO ISLEAR TLAGA GATE DRIVER S TAO NLSS IADO 1N L585 S DCA NLSS S RAL [SET 0’ OVERfLOH S TAO NNSI IADO 1N NSBS S 03A NNsa S 157 ’OAa IALL DONE. S JNI Ad IND. RETURN TOR NEXT SCAN S TAO NLSi ITES. SET SET TO AveaAae S SOL l.3A0 10 HITH L53 H000 S TAO NNSa S 0V! IOIVTDE 8 NEAR. 3000 [THE DIVISOR S TAO 34000 S CLL S DOA IJSTTK ISTOIE REHAXNDER S TOA l-0A0 nTVIDEND TNTO Ac S SNA IJNOeaaANOEO. S JNI T K11 IvEs. RESET S OOA vALJ INO. STORE S TAO VALU S 1A0 4 ummmmummmmummmm» (O‘DOIDQMUIS‘ «lawman» a§\\\\ 260 SNA [JVEQRANGEoo JNP I K11 [T§s. RESET CLA 3L IND. SJNTINUE TAO vALJ TAO :Aoco [:4AN35 T0 2.5 OOHPLTNsNT 3LL DCA IISTIK ZO'LSATIISTTK) Zz-P-OATIJSTI O50 ZONTTVUE 960 Irtxt-Iv)9sz.9s1.952 961 :ALL IVEVD 5 I962. 3;A :LL 8 J!” I 9537 £10