Simpson (Sam) Linke
Recollections, 1946 - 2005
The major portions of the following article appeared in the 2001 issue of Connections, the annual report and newsletter of the School of Electrical and Computer Engineering. I wrote it as a history of the development of computers in the school, both analog and digital, based on my recollections of involvement with both types and a little research to fill in some gaps. Although my major field of study is energy and power systems, I became interested in computer applications during graduate study at Cornell when I designed and built a small model of a power network calculator. The device was to be operated at 10 kHz to allow use of much smaller components than was the case for 60-hertz calculators. The objective was to promote construction of a full-scale10 kHz system for research and power-utility service. In 1953, however, Dean Hollister authorized purchase of a second-hand Westinghouse calculator that operated at 440 hertz and we were suddenly involved with a large analog computer. The rest is history as detailed below.
COMPUTER TECHNOLOGY IN THE ECE SCHOOL"I had my first class in Engineering Problems today. When I walked into the classroom the first thing I saw was a large white ruler hung along the top of the blackboard. It was about 10 feet long and covered with black scale markings. The top and bottom of the ruler were fixed and the third section in the middle could be moved back and forth. Our instructor said it was a teaching 'slide rule,' and when we got our own small versions they would help us tremendously throughout our engineering careers."
---Excerpt from a freshman engineer's first letter home, circa 1930.
In those days and well into the mid-1960s, engineering students were easily recognizable by the ever-present leather cases that contained their indispensable slide rules. EEs of the electric-power persuasion were particularly proud of their K&E "log-log-deci-trig with hyperbolic functions" rules that were designed to facilitate the solution of transmission-line problems. In the early 1970s when the versatile Hewlett-Packard and Texas Instruments hand-held calculators became readily available and relatively inexpensive, the slide rule era came to an end. It is possible that many of today's engineering students may never have seen one. The hand calculator is still a useful tool for routine calculations, but the personal computer is the centerpiece of the "digital revolution."
Computer technology first appeared in the EE School in 1950 when Professor Malcolm McIlroy '23, who had joined the faculty in 1947, established his pipeline network analyzer in the basement of Franklin Hall (now Tjaden Hall). This analog device consisted of an interconnected network of arrays of nonlinear resistances energized from a dc source (see Figure 1). A control center allowed the insertion of electromechanical measuring instruments in various points in the network as desired. The nonlinear resistances were tungsten filaments in evacuated glass envelopes similar to incandescent bulbs. Based on theoretical design as part of his Ph.D. thesis at MIT, Mac constructed various suitable lengths of the filaments, operated them at temperatures well below those of conventional light bulbs, and achieved units that would model the relationship in which pressure drop through a pipe is proportional to a fractional-power exponent of the flow rate. The idea behind this device was to take advantage of the very high sensitivity of tungsten resistivity to temperature, combined with the thermal-transfer characteristics (primarily convection and conduction) to obtain the desired mathematical function. Mac developed a quantitative design procedure for selecting the various physical parameters to achieve the proportionality constant and nonlinear behavior for a variety of physical pipe sizes. Since the device was able to simulate flow for water, natural gas, oil, and even traffic flow, a number of large-scale units were built for several utility companies around the country. In early tests, Mac studied the Cornell water-distribution network and discovered several weak points in the fire-protection system. Although accumulation of data was somewhat tedious, the device was excellent for providing instant visualization of incipient trouble spots in a system under study by means of the brilliance of the lamps in certain arrays in the network.
Figure 1. The McIlroy Pipeline Network Analyzer
In addition to conducting research on fluid flow, Mac acquainted the faculty with his analog device and urged several EE faculty members and Dean of Engineering S. C. Hollister to consider development of similar analogs for electric-power-network analysis. In 1953, an ac power-network calculator was purchased by the College of Engineering from the Westinghouse Electric Corporation and was installed in Phillips Hall when the EE School occupied the new building in 1955. The calculator was to provide system analysis service to electric-power utilities, serve as an instructional tool in power-system courses, and offer opportunities for power-system research. The relatively large-scale facility, shown in Figure 2, was installed in a large air-conditioned room on the second floor of Phillips Hall and consisted of panels filled with adjustable resistors, inductance coils and capacitors for representation of 128 transmission lines and 48 adjustable load circuits. Controlled transformers with adjustable voltage magnitudes and phase angles provided models of 18 generator units. Auxiliary plug-in panels allowed connections of these individual units into equivalent single-phase circuits to represent fairly large power-system networks. A separate control console and ancillary plotting table allowed insertion of electromechanical instrumentation to measure all required power-system parameters in any desired point in the network. The primary source for the calculator was a 25 kW, 440-hertz motor generator that was located in the basement of Phillips Hall. Design of the calculator to function at 440 hertz enabled the individual components to be physically smaller than would have been the case for 60-hertz operation.
Figure 2. The Cornell Power Network Calculator
Under the supervision of Professor Sam Linke, M.E.E. '49, with Larry Spencer, E.E. '34, as Chief operator and David Pulleyn as assistant operator, the facility provided professional system-analysis services to several electric utility companies and was used as a laboratory tool by a number of senior and graduate students in elective power-system analysis courses taught by Sam. When it became apparent in the early 60s that analytic studies would eventually be done by digital computer, the network calculator was decommissioned and course studies in load-flow and stability were transferred to the university computer mainframe. Power alumni of the period may recall those early analyses of the STEECO system (the fictitious Southern Tier Electric Energy Company) that were performed with punched cards. Eventually, the punched-card input was replaced by the now familiar keyboards and monitors.
Digital computers on campus at the time were represented by several International Business Machines card-programmed calculators (IBM/CPC) that were used primarily for accounting purposes. A CPC unit installed in Phillips Hall next door to the power-network calculator was designated as the Cornell Computing Center under the direction of Richard Lesser of the Department of Mathematics. The CPC unit was a noisy neighbor as it sorted thousands of IBM cards to perform its assigned tasks, but Dick Lesser assured the network-calculator operators that better machines were on the way and predicted that their big analog device would soon be obsolete. When Professor McIlroy used the CPC unit to compare a mathematical solution of a fluid-flow problem with his pipeline-analyzer solution to the same problem, he obtained essentially identical results. He found the CPC to be very slow but agreed with Lesser that an advanced digital computer would eventually supplant his analog device. In one of his last teaching assignments in 1956, before his untimely demise, Mac conducted a seminar, EE 4815, Fluid Network Analysis, that featured digital-computer methods.
The first of many major computer upgrades occurred in 1958 with the installation in Phillips Hall of an IBM 650, a machine with a high-speed magnetic-drum memory that could invert a 50 by 50 matrix in only 40 minutes (an incredible performance at the time). On one occasion the 650 was programmed to assist the power-network calculator in the solution of a power-system stability problem, and Ravi Sudan, then a visiting professor in the EE School, applied it in computations related to a study of a complex turbogenerator regulator problem. In 1960, the 650 was moved to Rand Hall, the new home of the Computing Center.
Computers in the EE curriculum were first mentioned in the Cornell Announcements of 1952-53 for an elective course EE 4123, Electronic Circuit Elements, followed in 1954 by a description of course EE 4810, Introduction to Electronic Computers. Both courses, taught by Assistant Professor Al Jackson, were concerned principally with analog computers applied to the study of control systems. Al later designed and built the Cornell Electronic Analog Computer (COREAC) a small working unit based on these principles. (See Figure 3).
Figure 3. The first Cornell Electronic Analog Computer (COREAC)
In the early 1960s, computer courses in the EE School gradually moved from analog to digital mode. Associate Professor Norman Vrana, M.E.E. '51, taught EE 4810 as a course in analog computers for several terms before beginning his long-term concentration in the School on courses in digital-computer design. In 1961, Assistant Professor H.C. Torng, M.S., '58, Ph.D. '60, introduced course EE 4820, Switching Theory and Digital Computers, which was described in the Cornell Announcements of that year as a course in "switching devices, logical formulation and realization of combinational switching circuits, number representation and codes, simple memory devices, counters, shift registers, and arithmetical units in a digital computer."
By 1963, H.C., by then an associate professor, added course EE 4821, Switching Systems, which offered an integrated study of switching systems including general-purpose digital computers and an introduction to the general theory of learning machines. In the following year, H.C. changed the two courses to EE 4587-88, Switching Systems I and II, thereby firmly establishing digital computers as an elective area of study in the EE curriculum.
In 1964, the first mention of the Cornell Computing Center appeared in Announcements, with a description of the computer facilities available for use by students and faculty, including a Burroughs 220 digital computer and a Control Data 1604 and 160A systems. The IBM 650, initially housed in Rand Hall, had been replaced with these units, and they, in turn, were superseded in 1966 by more advanced systems installed in Langmuir Laboratory at the Tompkins County Airport. Direct lines to this "mainframe" were made available to remote computer terminals in several buildings on campus.
PROGRESS IN COMPUTER STUDY
The remote terminal that was used primarily by engineering faculty and students was housed in Room B7 Upson Hall complete with IBM-card readers, keyboard card punchers, and card-to-paper printers. Although those facilities were primitive by today's standards, their presence allowed the introduction of direct use of computers in the engineering college.
In 1966, under the new Division of Basic Studies (DBS), all engineering freshmen were required to take course Eng. 104, Introduction to Engineering, which included an introduction to digital computing, use of the Cornell computing language CORC, and computer applications.
In 1968, the language was changed to CUPL until a revised DBS sequence of courses was adopted and the course became ENGR 105, Elements of Engineering Communication, with FORTRAN as the official language. In 1972, when the Department of Computer Science began to offer CS 100, Introduction to PL/1, in which problems assigned to be programmed in that language were processed on the computer, the first eight weeks of ENGR 105 were made identical to CS 105.
By 1978, a new course, DBS 105, Introduction to Computer Programming, was entirely identical to CS 105. In the same year, ELE 230, Introduction to Digital Systems, became the first required computer course in the upper class EE curriculum. Alumni of the period will recall the computer "gridlock " that resulted from the massive use of the remote terminal by the students in those courses.
In the 1970s, the field of digital computing became a definite presence in the EE School. When the power-network calculator in the EE School was decommissioned, Sam Linke, and his students used the Upson Hall terminal to analyze electric-power systems on the mainframe computer. Because of the heavy computer traffic, the turnaround time for the solution of a problem was usually about 24 hours unless students chose to appear at the terminals in the early-morning hours. Control courses in the School also made use of the mainframe with similar delays. Several EE elective courses in computer engineering became well established in this period as the earlier courses began to concentrate on digital techniques. Norm Vrana taught EE 674, Analog and Hybrid Computation and EE 676, Computer Structures, and H.C. Torng taught EE 675, Switching Circuits and Logic Design, and EE 677-78, Computer Architecture and Design I and II.
In 1975, Associate Professor Chris Pottle, one of the founders of the Cornell Department of Computer Science, who had previously developed a widely used computer language called CORNAP, introduced EE 624, Computer Methods in Electrical Engineering, which explored techniques for solving electrical engineering problems on the digital computer.
At the end of the decade Cornell Announcements reported that the university central computing facilities now included an IBM 370/168 system, a Digital Equipment Corporation (DEC) System 2060, and "two recently installed data-communication networks (TELENET and TYMNET) that give Cornell computer users access to computing facilities in 40 states as well as Mexico, Canada, and Europe," a clear indication of things to come. In the remote terminal in B7 Upson the IBM-card era finally came to an end with the installation of relatively modern workstations.
Coincident with the development of the computer-engineering curriculum in the EE School, increased dependence on computers for both instruction and research suggested a need for local computing facilities. In the early 1970s, the systems group installed a then top-of-the-line PDP11/40 minicomputer in Phillips Hall and used it productively for several years (See Figure 4).
Figure 4. Students operating the PDP 11/40 digital minicomputer
Faculty and student interest in power system engineering increased following the historic 1965 Blackout in New York City. Responding to the apparent need for advanced techniques for control and protection of the expanding power-system networks and to the potential for the application of digital computers to power-system operation and control, professors James S. Thorp '58, Christopher Pottle, and Robert J. Thomas applied their expertise to problems in the power field and encouraged graduate students to undertake thesis research in the discipline. Together with Sam Linke, Joe Rosson, M.E.E. '51, and other members of the EE faculty with related interests, the Cornell Program in Power Systems was formed and research proposals were submitted to suitable energy agencies. The program was enhanced by the receipt of a $250,000 grant from the Kettering Fund for the establishment in 1976 of a power-system laboratory in Phillips Hall to be known as the Eugene W. Kettering Energy Systems Laboratory (Figure 5), in memory of the late Gene Kettering, E.E. '30. The principal component of the new laboratory was a transient network analyzer (TNA) that was designed by members of the power group and constructed for the most part by Charles R. Strohman '77. The TNA was augmented and controlled by a VAX 750 digital computer that could be accessed in the laboratory proper and also by remote terminals in Phillips Hall. This latter feature was of great benefit to the students who were studying the STEECO system since instant access and instant turnaround removed the tedium and delays associated with the university mainframe computer. Students could now insert their data using screen monitors and keyboards and obtain virtually instant turnaround from the printer in the Kettering Lab, a procedure that was a precursor of today's commonplace internet usage. In recent years the TNA has been relocated in Rhodes Hall (the Theory Center Building) and the now obsolete VAX 750 has been replaced by several workstations connected to the Phillips Hall computer network. In a manner reminiscent of the old power network calculator days, the TNA has been used occasionally by several commercial companies for special relay studies.
Figure 5. A portion of the Kettering TNA
In 1983, the School acquired a Data General Corporation MV 8000 computer that supported workstations throughout the building to form the first computer network in Phillips Hall (See Figure 6). The following year saw the installation of a dozen Macintosh personal computers (PCs) fitted with a logic-works program to form a local-area network (LAN) for the required course EE 230. The former strictly laboratory format was replaced with a simulation of logic design capability that was augmented afterwards by the addition of programmed logic devices. Students could then design logic circuits and obtain output devices in the form of real chips that could be tested in real circuits.
Figure 6. The first Phillips Hall Network computer MV 8000
By the end of the 1980s, computer engineering had become a well-established area of study in the EE School. In addition, basic engineering studies, now known as the Office of Undergraduate Affairs, recognized the importance of computers in all engineering studies by adding to the computer requirement of ENGR 100 at least one additional course with a significant amount of computer application, either as a distribution course from the "scientific computing" list or as part of a field program. The computer language in use at the end of the decade had become PASCAL, to be followed after several years by the C language.
At that time, JAVA was the language of choice together with some instruction in METLAB. EE 308, Fundamentals of Computer Engineering, was taught by Assistant Professor Miriam E. Leeser, B.S.E.E. '80, as an EE core requirement. Norm Vrana's highly popular elective courses, EE 475, Computer Structures, and EE 476, Microprocessor Systems, were in full swing, H.C. Torng introduced two new courses, EE 545 and EE 546, Computer Networks and Telecommunications I and II, and Chris Pottle offered a new course, EE 576, Parallel Processing, the study of architectures related to the development of "supercomputers," machines designed to provide high computation rates for large scientific problems. Associate Professor Tony Reeves introduced an interesting new computer area with course EE 547, Computer Vision. Not surprisingly, the 1986-87 issue of Cornell Announcements listed computer engineering for the first time as an "area of concentration" in the field of electrical engineering with 14 required or elective courses.
COMPUTER STUDY MATURES
During the past 15 years, the presence of newly appointed computer-oriented faculty members and the acquisition of high-quality computer facilities have created major changes in electrical engineering education in the school. In 1991, Director Noel MacDonald organized the effort to allocate and renovate three third-floor classrooms in Phillips Hall for centralized classroom computer usage. Under the direction of Professor Emeritus Norm Vrana, one room was installed with Macintosh PCs, another with Hewlett-Packard (HP) Vectra PCs and the third with HP workstations. This facility made it possible for any EE course to employ computers directly in classroom activity or labs as well as for outside class assignments. Over the years, the original workstations have been upgraded regularly as new equipment became available. EE faculty members in diverse fields have taken advantage of the expanded facilities to integrate computer techniques into their design and laboratory courses and require students to use the workstations for the solution of assigned problems and the development of projects. This trend toward computer-aided instruction was further enhanced two years later in the EE School by the installation of the undergraduate teaching laboratory that makes use of leading-edge technology and instrumentation and incorporates state-of-the-art practices found in leading electrical and electronics industries. Through connections to the Phillips Hall computer network (EENet), each component of the laboratory has remote access to the university computing center now located in the Theory Center in Rhodes Hall. In 2000, the Intel Computer Systems Teaching Laboratory was established to support both teaching and research in computer systems. During this period of expansion, state-of-the-art computer equipment was brought to our laboratories through the efforts of Emeritus Professors Norm Vrana and H. C. Torng who negotiated generous grants of computer hardware and software from Hewlett-Packard, Intel, Apple, Altera, Sun, Digital Equipment, and Motorola Corporations. More recently in 2004, EE grads Stephen G. Meier, '89, and Abeezer S. Tapia '02, arranged for a donation of 40 top-of-the-line computers by Advanced Micro Devices that satisfied an urgent need for high-performance facilities in the ECE School.
An examination of the advanced courses now offered in the school reflects the profound influence of computer technology on the curriculum. Graduate and undergraduate students who wish to concentrate on computer engineering may choose their electives from a wide variety of available subjects and can augment their studies with appropriate courses offered by the Department of Computer Science. Students with interests in other EE disciplines can elect computer courses that will acquaint them with the impact of computer technology on their particular fields of choice. ELE 476, Digital System Design Using Microcontrollers, is a popular course in this category taught by Senior Research Associate Bruce Land, Ph.D. '76, (Neurobiology & Behavior and Electrical Engineering). Students working in pairs design, debug, and construct several small systems that illustrate and employ techniques of digital system design acquired in previous courses (see Figure 7).
Figure 7. Students in Dr. Bruce Land's digital-design lab concentrate on their projects.
As the now properly named School of Electrical and Computer Engineering enters the twenty-first century, faculty members and their graduate and undergraduate students all benefit from the availability of these creative techniques and powerful tools that facilitate the study of fundamental principles in all areas of electrical and computer engineering.
Electrical and Computer Engineering
-------June 1, 2006
University Library Rare and Manuscript Collections for Figures 1-6;
Bruce Land for Figure 7.