The Dawn of the Programmable Logic Controller (PLC)

Summary

Part 1 in a Series about the History of the PLC

This article was originally published in the Spring 2015 Edition of PULSE.

By Ken Ball

Programmable Logic Controllers (PLCs) were more of a technology evolution than a radical development.  A noteworthy aspect about the origin was that quite similar concepts were pursued by three separate companies—nearly concurrently but unknown to each other.

The three companies were Bedford Associates in Bedford, MA; Struthers-Dunn Systems Division in Bettandorf, IA; and General Motors Hydra-matic Division in Yipsilanti, MI. 

Bedford and Struthers-Dunn were automation suppliers whereas Hydra-matic produced automotive transmissions.  Hydra-matic had a strong interest in developing an industrial controller to improve production operations.  To this end, Hydra-matic worked and shared information with two computer technology companies - Digital Equipment Corporation (DEC) and Information Instruments Incorporated (3-I).

DEC and 3-I built prototype units for Hydra-matic so overall five companies were involved, along the three separate paths, in developing a digitally based, easily programmed industrial controller.  With each of these three paths PLC concepts, prototype building & testing, and factory evaluations occurred over the 1967-to-1971 time frame. 

The timing was not random.  Rapid advances in solid state electronics, computers, and digital signal handling in the early-to-mid 1960s provided the technology base for the PLC.   The Figure 1 Chart provides a vivid example of the profound changes solid state electronics made in simple flip-flop circuit characteristics.

Figure 1

Any review of the history of the PLC uncovers the profound advances made over the decade of the 1960s—probably the most dynamic decade of the 20^th century.  Some examples of progress in the ‘60s are listed for a more comprehensive historical perspective.

• From printed circuit boards with shaky $20 transistors and wire wrapped terminals to ICs and then highly reliable LSICs. 
• Numerical control (N/C)—though the Europeans and Japanese took the lead. Migration to CNCs began in the late ‘60s.                
• Transistorized Op Amps and Analog Computers  (Analog Devices was formed in 1964)
• Better SCRs and triacs greatly improved motors motor drives and motion control. 
• Video Display Terminals developed by Bell, MIT, & GE.  Early CAD systems.  
• Lasers, Industrial Robots, and Automatic Identification Systems were introduced.

Another line of 1960 era products which related to eventual PLC functionality were called Program Controllers.   These were mostly timing and sequencing devices used for simple systems operation or as components in more complex systems.  These devices had a sufficient market to attract the attention of the trade press.   The September 1970 issue of Instruments & Control Systems magazine carried a survey listing about 60 such devices. A segment of the listing is shown in Figure 2. 

Even though PLCs inherently obviated the need for many such devices, it is interesting that only five of the listed suppliers developed a product line of PLCs.  (Something akin to vacuum tube companies not producing transistors).   The Figure 2 segment shows two of the eventual PLC supplier companies - Industrial Solid State and Industrial Timer.   Honeywell also became a PLC supplier through a marketing agreement with ISSC.   

Figure 2

Emergence of Modicon

Founded in 1964 by Dick Morley and George Schwenk, Bedford Associates was a New England control systems engineering company.  Their primary clients were New England machine tool builders who were facing increasing competition from European and Japanese machine tool suppliers; in addition to their general needs to improve machine tool performance and N/C options.  Also, there was high interest among machine builders and their customers to leverage the benefits that new minicomputers could provide for high throughput manufacturing lines.

After several machine tool system projects utilizing minicomputer controls, Morley became exasperated with the usual 6-months programming and debugging that was required to start systems up.  By 1967, he began “Imagineering” ways to build a much more user-friendly, computer-like control unit which could be easily programmed and re-programmed.   The Bedford staff was well experienced in factory operations and they had worked in harsh factory environments.

According to Morley, his thoughts gelled on New Year’s Day 1968 while walking off the effects of the prior night’s New Years' eve party.  Indeed, his 12-page concept memo is dated January 1, 1968.  The Bedford group had a prototype up and running by March and in April the new unit--which they had nicknamed Stupid--was shown and described in a demonstration at Landis Machine.  The demonstration was well received and the new controller was subsequently shown to Bryant Grinders and several other potential clients.   A major result of these contacts and feedback from machine builders was the decision to use ladder logic for programming. 

In May ’68, Bedford’s Sales Manager Lee Rousseau attended the annual Westinghouse Machine Tool Forum in Philadelphia.  GM Hydra-matic’s Bill Stone was a speaker and presented the Hydra-matic Division’s concept for a Standard Machine Controller, which included desired performance specifications.  The Bedford prototype factory controller well meshed with Stone’s envisioned standard controller concept.

Rosseau was beside himself in the audience mumbling “we already have one of these” and he made plans to visit the Hydra-matic plant as soon as possible.  Bedford Associates became the seventh controls company given the Standard Machine Controller specifications. Visits to the Detroit area completely revised Rosseau’s market perceptions and a separate manufacturing operation was proposed.  Bedford Associates would remain an engineering services provider and the new manufacturing company was named Modicon—an acronym of MOdular DIgital CONtroller.    

Developments at Struthers-Dunn

As a major relay supplier, Struthers-Dunn (S-D) was well aware of chronic problems associated with relay based controls in high volume automotive production operations.  Along with a wide range of relay products, S-D also enjoyed the prestige of having supplied many control systems engineers and technicians with their pocket-size handbook: Relay Engineering Guide.  The guide taught relay logic and diagrammed clever relay applications.

S-D had established a Systems Division at Bettendorf, Iowa to develop solid state logic devices and digital controls.  In 1967, the Division hired Pete Bartlett, an accomplished EE who had been developing diode matrix circuitry for Eagle Signal.  Bartlett began developing a dedicated industrial logic computer. In 1968, he was in touch with Ingersoll Engineers relating to a Ford production line application.

Two of Bartlett’s engineering colleagues followed him to S-D Systems.  Eagle Signal’s parent company (Gulf + Western) initiated trade secret legal action. This enjoined Bartlett’s new group from further development until the litigation was resolved—roughly a year later.

In the interim, Ingersoll turned to DEC to supply their production line controller needs.  After the litigation delay, Bartlett’s group resumed their “Industrial Logic Computer” development. They were able to build a prototype unit—named the VIP--in time for display at the 1970

Machine Tool Show.  In 1971 a VIP was installed for a beta site evaluation at the Ford Chicago Heights Stamping Plant and operated for several years.  Sales of the VIP began in 1972.

General Motors Hydra-matic Standard Machine Controller Project

In the late 1960s, GM’s Hydra-matic Division at Ypsilanti, MI was producing upwards of 2 million transmissions a year.  Long, complex transfer lines were used to machine and assemble parts.   

One such line built to convert raw castings into finished transmission cases was 420 feet long, had 58 work stations, and 329 cutting tools. 

Control of transfer lines were by corresponding lines of control cabinets (usually a cabinet for each machine) housing relays, motor starters, timers, sequencers and the like.  Handling high voltage line power, excessive heat, arcing, e/m noise, along with limited relay and component reliabilities often resulted in control component malfunctions.   Downtime on one line was estimated to cost $1600/min ($96,000 per hour)—in 1965 dollars when new cars sold for about $5,000.

Like most of the automotive industry, GM invested heavily in machinery and equipment upgrades to improve quality and productivity.  For much better production monitoring and manufacturing data handling, Hydra-matic had installed three minicomputer systems between 1965 and 1969 and a fourth was planned.  Two of these were IBM 1800s. The first monitored 24 dynamometer test stands and the second monitored a transmission case assembly line.  

The third mini was a Varian 620i used to control a forward clutch assembly machine.  Needed interfacing electronics were being developed by Information Instruments, a long term Hydra-matic contractor.   John Dute, 3-I President, personally worked on the interfacing and, in his opinion, this project was the world’s first computer controlled assembly operation. 

The minicomputer systems operated successfully and Hydra-matic management justified costs with documented decreases in rejects, better troubleshooting, less scrap, and the like.  However, relay panel controls along transfer lines remained major maintenance headaches.  Relays themselves consumed power and generated heat; contacts began to bounce and control component short circuits created havoc.  The entire Hydra-matic staff, from operators and floor electricians to top management, wanted to rid their world of relay panels.

A number of the Division’s engineers from several groups were exchanging ideas and pushing concepts for a small, simple machine controller which would replace relays.  Key persons included Test Supervisor Ed O’Connell, Systems Engineer Jim Bevier, and Control Engineers Dave Emmett and Len Radianoff.   A project was authorized under the Process Control Group, headed by Ed Clark.  The staff’s controller interests were shared with John Dute of 3-I and DEC Representative John Dumser.   Both offered ideas and suggestions.  Reportedly, the GM staff leaned towards DEC--with their stature and minicomputer track record--as a preferred supplier of the envisioned machine controller.

In March of ’68, Jim Bevier prepared sketches and diagrams for an I/O structure for a Standard controller.  In April, Dave Emmitt suggested procuring and applying the specified Standard Machine Controller.  The proposal was endorsed by Bill Stone, Supervisor of the Division’s Machine & Gear Development.  Later in May, Stone presented a paper describing the Standard Machine Controller concept and preliminary specifications at the annual Westinghouse Machine Tool Forum in Philadelphia.  

Key items in Hydra-matic’s Standard Machine Controller specifications included the following:

• The unit will be of modular construction and operate in a factory environment subject to nearby high voltages and vibrations.  Built-in isolation would accept 120 vac digital signals and would provide at least sixteen 120 vac, 4 amp outputs.
• Using solid state components, it would have 32 inputs expandable to 256 and 16 outputs expandable to 128.  Stored information or programs would not be altered or lost due to a system power failure of up to 12 hours.
• It would be easily programmed and re-programmed.  It will have at least 1K of memory expandable to 4K.
• It would be capable of handling eight simultaneous timing functions with a timing range adjustable from 0.1 to 10 sec.

A final form of the solicited “Standard Machine Controller” was completed in early June and was given to four vendors--Allen-Bradley, DEC, 3-I, and Century Detroit.  Later on, Cutler-Hammer, Cincinnati Milling Machine, and Bedford Associates obtained copies.  Out of this group of seven, only three--DEC, 3-I, and Bedford Associates--delivered prototype controllers for evaluation.

Specification reviews, board and hardware designs, proposal preparations, building and testing took the best part of a year.  The first unit delivered in June, 1969 was the DEC PDP-14 which was installed to control a gear grinding machine.  Later in the summer, 3-I delivered their unit designated the PDQ-II.   PDQ stood for Program Data Quantizer and the 3-I unit was installed to control a segment of an assembly machine.   Last in was Bedford Associates—now known as Modicon.  Delivered in November, their unit was Bedford’s 84^th project of 1969 and so carried the designation the 084.  Like the DEC unit, iwas installed to replace a relay panel controlling a gear grinding machine.        

The prototype controllers performed well and interest was spreading both inside General Motors and to automation supplier and manufacturing engineering services companies.

Being a prototype supplier and Hydra-matic contractor, 3-I President John Dute had perhaps the most strategic view of the market potential for the new controllers.  Dute realized his small company lacked resources needed to be a serious competitor and eventually Allen-Bradley (A-B) expressed interest in providing needed support.  Allen-Bradley bought 25 % of 3-I, rights to market the PDQ-II controller, and an option to acquire 3-I.  A-B exercised their option and acquired 3-I in late 1969.     

Performance & Publicity

According to Hydra-matic management, all three prototype units met the specifications and remained running into 1970.  In May, Bill Stone presented a follow-up paper at the 1970 Annual Westinghouse Machine Tool Forum held in Pittsburgh.  The presentation on the new controller was based almost entirely on the DEC PDP-14 unit since it had been in operation for nearly a year.

Stone’s slides showed that for every 100 machine inputs, their relay controls averaged 26 control failures per month.  Early use of the controller reduced failures by about two-thirds—to a projected 8 failures per month for every 100 machine inputs.  In a comparison of monthly maintenance hours for electrical repair, transfer line relay controls required roughly 26 hours per month for every 100 machine inputs.  With the Standard machine controller, required electrical maintenance hours were cut in half—to 13 hours per month per 100 machine inputs.

By this time many automotive manufacturing engineering and management persons were becoming familiar with the standard controller project—by industry grapevines as well as from a number of  technical paper presentations.  Interests ran especially high among manufacturing engineering companies and automation suppliers.  GE and Square D quickly made marketing arrangements with Modicon and DEC, respectively.  Square D gained the rights to market the DEC PDP-14 and Modicon private labeled the 084 as GE PC-45 Controller. 

Indicative of the attention the Standard Machine Controller was gaining in the automotive industry, seven companies reacted quickly enough to display the new PLCs at the 1970 Machine Tool Show.  This occurred only months after Bill Stone’s Forum presentation and preliminary

Hydra-matic plant beta-site evaluations.  Also, Struthers-Dunn Systems had resumed their R&D project and displayed a VIP unit at the show.  The companies and corresponding products displayed included:

Performance evaluations continued at the Hydra-matic plant throughout 1970 and into 1971.

With continued operating experience, the Modicon 084 became the controller of choice for factory engineers and electricians.  First and foremost, Modicon programmed in ladder logic, whereas the PDQ-II and PDP-14 programs were written in Boolean.  This was fine for computer engineers but average plant engineers and electricians understood ladder and easily adapted to the 084.  Other PLC builders soon emulated Modicon and offered ladder logic.

084 programming was relatively straight forward.  The user plugged in a programming unit, selected an appropriate software module and keyed in ladder diagrams.  Modicon also established a phone service for users programming and troubleshooting support.  In an early interview based on a Bedford machine tool project, Dick Morley stated that their new PLC had reduced programming time from 6 months to 6 days. 

In contrast, to re-program the PDQ-II, a Boolean program was written to punch a paper tape from a teletypewriter using a minicomputer interface.  A special loader was used to load the program.   The PDP-14 also required a Boolean program on punched tape and used a woven wire rope circuit board memory.  A new program, tape and memory board was sent back to DEC for the program change—with about a one week turn-around. 

The PDQ-II and PDP-14 were replaced in 1971.  The 084 remained in service for more than 10 years.   Hydra-matic eventually returned the initial 084 to Modicon and it has been on display at the Smithsonian.  A similar 084 is shown in Figure 3 with several of the original Modicon staff.

Figure 3 - PLC Pioneers, from left to right, Dick Morley, Tom Bossevain, George Schwenk, and Jonas Landau