First-Hand:Six Decades of Calculations

Six Decades of Calculations

Introduction

This is a first-hand history of all kinds of computing devices I used over 62 years. It started (with a slide rule) at M.I.T. in 1946, then Georgia Tech in the late 1950s, S.M.U. in the 1960s, as an employee of four defense electronics companies, and continues in my retirement through 2009. I am a user and not a designer of calculators and computers, and descriptions of most equipment I mention are readily found by Google search or at www.wikipedia.org.

My three EE degrees have enabled design and synthesis of electrical filters and impedance-matching RF networks in frequency ranges from VLF through K band in conjunction with numerical methods and analysis, especially optimization (nonlinear programming), and computer programming in FORTRAN, BASICA, QuickBASIC, Visual BASIC, and C languages. Many of these calculations have been documented in my three books, which can be downloaded in searchable PDF files (click on citations in blue type):

Circuit Design Using PCs here, Optimization Using PCs here, and Broadband Direct-Coupled and Matching RF Networks here.

Availability or absence of computing equipment due to management decisions is emphasized, especially my passion for autonomy for the working engineer. Obviously, mainframe computer systems can pull a heavy load, but I have one supreme thought in that regard: Thank goodness Paul Revere was not stuck with a team of mules!

Genesis

Two mandatory purchases by a student entering M.I.T. in 1946 were a K&E Model 4083-3 Hyperbolic Log Log Duplex wooden slide rule and a time-honored handbook with 95 pages of formulas and 172 pages of tables, especially logarithms (Burington, 1940). (References are listed in the last section.) Talk of Vannevar Bush’s 1931 analog differential analyzer at M.I.T., outdated by the first digital computer, ENIAC announced in 1946, was just motivation, because we had to use that slide rule and tables for all our calculations. The only different calculations I saw were by about a half-dozen women eternally employed in the basement of the Aeronautics building, passing around algorithm data sheets that were processed with hand-driven mechanical calculating machines.

Threatened with draft into the Army in 1949 and not enamored with over two years of Army ROTC at M.I.T., I entered Navy flight training and became a Navy Pilot and Electronics Officer for the next ten years. The only Navy computing aids were simple circular slide rules and a pencil-driven lap vector-graphics board to answer the question: Where is the aircraft carrier and how do I get there? I went back to college at Georgia Tech in 1956, with my trusty slide rule. However, I became deeply interested in passive circuit theory, but it was terribly obvious that tedious slide rule calculations misplaced the decimal point, were grossly inaccurate, and were definitely not fun.

Mainframes

I was about to graduate Georgia Tech in late 1958 when I heard that the Southern Railroad in Atlanta had one of the first IBM mainframe digital computers, and I even considered interviewing the railroad for that reason alone. But having operated and supervised maintenance of many Collins Radio transmitters and receivers in the Navy, I was hired into their Cedar Rapids, Iowa, plant in January 1959. I was delighted to find they also had one of the IBM 650 computers, a rotating memory-drum vacuum-tube system with only punch-card I/O and a numerical coding scheme similar to an abacus. The 650 could be programmed in assembly language and at a higher level with SOAP, the symbolic assembly program. That was not for engineers, but Collins mathematician Dr. Victor Wayne Bolie wrote an interpreter for the 650 that was perfect for engineering calculations. So, within a few weeks, I was drawing flowcharts on wide computer paper laid out many feet on the floor to program RF circuit design answers to about seven figure accuracy, leaving other engineers with slide rules to claim that those narrowband circuits would not work – but they did!

Collins moved my family to Richardson, Texas, in North Dallas after only four months in Cedar Rapids, but with no Collins IBM 650 computers in the Dallas area. However, geological exploration company Core Laboratories in the downtown Dallas industrial district let us use their 650 from 3 AM to 7 AM. It sounds crazy, but several times a week I took several Collins engineers and another employee who could program the 650 plug board (to read and print punch card data) as well as a keypunch operator from the Garland suburb to get circuit design answers during the wee hours at Core Labs. Months later, Collins Richardson plant installed an IBM 650 computer, but the beancounters owned it and engineers were badly served. For example, our engineering department was not allowed to own or use an IBM 026 card keypunch machine, those being locked up in a room with women operators and verifiers. Typically, we filled out and submitted data sheets, the resulting cards then were processed offline overnight, and we found the output (often with several maddening errors) the next morning. Later we could also get graphs, but they were put in the company mail for another big delay. A few computing sessions were conducted at Ling-Temco-Vought company in Garland on an IBM 7070 (transistorized 650) and on a very fast CDC-6600 at S.M.U. in Dallas.

Collins Richardson Division bought a Univac 1108 mainframe in the mid 1960s, and engineers were allowed access by remote CRT terminals. I finally talked our division financial officer (later Rockwell International CFO) into buying one TI portable (really heavy) remote terminal that included a thermal printer. He had badgered me to tell him exactly what the percent usage would be and other nonsense. Getting engineering answers was like pulling teeth.

Anything But Mainframes

In the 1970s I first had a Monroe LN non-electric mechanical calculator, then several Friden electromechanical models and finally a Friden EC-130 electronic calculator. Meanwhile, Collins Dallas engineers tried to get the company interested in buying a new IBM 1604 computer or even the first IBM desktop machine using the APL language, the 5100. We also proposed leasing DEC PDP, VAX and Wang minicomputers with no better response.

The computing tide for engineers began to turn with the handheld HP-35 scientific calculator in 1972, followed by its programmable cousins HP-65 and HP-67/97 in 1974 and 1976, respectively. Programming was in reverse-Polish notation (RPN), which appealed to many engineers. Practical circuit software soon followed (Murdock, 1979), (Cuthbert, 1983). Affordable HP desktop calculators HP-9815 (RPN) and HP-85 (HP BASIC) were available in 1975 and 1979, respectively, and both included an IEEE Bus controller for compatible instrumentation. Naturally, I personally bought every one of these handheld and desktop computers as soon as they were available. Incidentally, I left Collins Radio for Texas Instruments just down the road in 1972, returning to then Rockwell Collins in 1974. I used the TI mainframe computers successfully, but clung to my RPN HP-35 calculator, much to the chagrin of TIers and their algebraic handheld calculators.

Personal Computers

The MITS Altair 8800 sparked the PC revolution in 1975, and although it was programmed in Altair BASIC it was clearly too primitive for serious engineering design work. However, the Apple II and the Commodore PET in 1977 had the required potential. I immediately bought the 8K (versus 4K) RAM PET model and used it both to write my Circuit Design book and to program the included calculations: “I wrote this book to show how effective personal computers can be in circuit design.” (Cuthbert, 1983:vii). Appendix A listed 17 HP-67/97 calculator programs and Appendix B listed 25 PET BASIC programs. (The HP-67/97 programs were later made available in BASICA.) Dot matrix printers were readily available, but high quality printing was a problem. I first tried converting a new IBM Selectric ball typewriter to be a printer but later bought a Diablo 630 wheel printer that performed satisfactorily.

The IBM PC in 1981 was the real breakthrough. It was based on the Intel 8088 chip set, which was improved by addition of the 8087 math coprocessor. That coprocessor increased performance by several hundred percent and provided extraordinary precision for challenging circuit design computations. I soon talked Rockwell Collins Dallas corporate management into a payroll deduction plan for engineers to buy IBM PCs. The loan balance soon exceeded $1 million, and then other company professionals began to feel left out, so the program was terminated after a year or so.

At one point our engineers were so enthusiastically involved that a member of the Chief Engineer’s staff was told by his boss to stop spending all his time playing with those PC toys. Actually, Corporate did design a Rockwell PC and I was strongly invited to try it and push it. Frankly, it was inferior and fortunately faded away. That same Chief Engineer invited me to downtown Dallas in 1981 to look at a demonstration of the Xerox Star computer system, which introduced the graphical user interface (GUI) and most features common to modern PCs. I had been satisfied with the command-line interface of the Univac 1108 mainframe and the IBM PC, so I told our Chief Engineer that the GUI was cute but an unnecessary waste of limited CPU power. Of course, microprocessors then have been eclipsed by those now, so I do appreciate today’s GUIs.

Languages

After Vic Bolie’s Interpreter for the IBM 650 in 1959, I learned FORTRAN II in a class at a Dallas IBM office. That has served well on both mainframes and on PCs loaded with Microsoft and other FORTRAN interpreter/compilers. Also, early extensive experience with the necessarily compact RPN code improved my programming skills. Knowing FORTRAN simplified programming in the variations of GW-BASIC, BASICA, and QuickBASIC. I worked at E-Systems in Dallas after retiring from Rockwell Collins in 1987. I reported to another Ph.D. who insisted that I program in the C language. Fortunately, I found a C compiler and also mimicked FORTRAN as much as possible. Nevertheless, I did not like pointers to pointers and other tricks in the C language.

I note the modern trend in high-level languages such as Mathcad, Mathematica, and MATLAB. Many young engineers use these tools to great advantage, but I have decided not to invest in a new trade this late in life.

The first spreadsheets, Lotus 1-2-3 in 1983, VisiCalc in 1987, and Quattro Pro in 1988 were very clever and I used these on many PCs. I have since become a great admirer of Microsoft EXCEL, mainly because of the graphical displays, SOLVER optimizer, and integrated Visual Basic for Applications (VBA). Web site http://members.cox.net/trcpep/services.html offers download of PDF files describing two EXCEL programs that utilize all the above advanced features.

Retirement Tools

My present domestic stable includes two Apple Mac Minis with HP LaserJet printers. My office has an HP-m7680n Media Center (XP OS) and an Apple iMac (Leopard OS), each operating an Intel Core 2 Duo processor. I have an HP-48GX programmable calculator I never use and a cheap HP-30S scientific calculator I use regularly. I have two HP laser printers and one HP Deskjet color printer.

My broadband Internet portal is connected to an Apple Airport wireless router that serves all computers as well as an iPhone and Apple TV. I freely admit having opposed Apple products from their beginning until a few years ago when two factors dominated: Apple computers are essentially immune to virus threats, and the Apple operating systems (OS) greatly outperform Microsoft offerings. Of course, my huge Microsoft software library and my own programs mandate having a Microsoft computer available. Fortunately, Cox broadband service includes anti-virus protection, annoying as that is.

References

Burington, R. S. (1940). Handbook of Mathematical Tables and Formulas, 2nd Ed. Sandusky, OH: Handbook Publishers.

(Cuthbert, T. R. (1983). Circuit Design Using Personal Computers. NY: John Wiley. Also, Melbourne, FL: Krieger Publishing Co. (1994).

(Cuthbert, T. R. (1987). Optimization Using Personal Computers with Applications to Electrical Networks. NY: John Wiley.

(Cuthbert, T. R. (1999). Broadband Direct-Coupled and Matching RF Networks. Greenwood, AR.: TRCPEP Publications.

Murdock, B. K. (1979). Handbook of Electronic Design and Analysis Procedures Using Programmable Calculators. NY: Van Nostrand Reinhold.