The History of Compucorp
By Rick Bensene
Major Revision 10/31/2003
Calculators made by Compucorp are extremely interesting and wonderful calculating machines. Compucorp machines had a real style about them, both in terms of their physical appearance, and their high-end functionality. Technologically, Compucorp's calculators (and those sold by OEM customers including Monroe(US), Smith-Corona/Marchant [SCM] (US), Deitzgen(US), Sumlock(UK), IME(Italy), Seiko(Japan) and Ricoh(Japan)). were highly advanced, pushing the state of the art in large-scale integrated circuit technology, as well as utilizing a very computer-like architecture. Compucorp's unique architecture used a multi-chip general purpose CPU (the precursor to a microprocessor) with specialized ROM-based software making up the personality of each different model of calculator. The flexibility of the architecture allowed Compucorp to manufacture a wide range of calculators for many different types of applications, including scientific, statistical, financial, and even for surveying work. Compucorp's later calculators blurred the line between calculator and computer, offering highly-expandable calculator processors that could be connected to a diverse range of peripheral devices to form what, for all intents and purposes, was a small computer system.
Compucorp got its start the way that many high-tech companies get started -- as a spin-off. Compucorp actually had its genesis in another company, Wyle Laboratories, of El Segundo, California. In 1961, Wyle Laboratories acquired a small company specializing in digital logic modules and custom electronic systems. This company had two consultants named Thomas Scuitto (an electronics designer, who designed the calculator) and Matthew Alexander (who was the business side of the team), that were working for them developing an electronic calculator. Since the consultants were part of the team now, the calculator effort simply migrated to Wyle Labs. A substantial amount of development had already been done on the calculator project, enough that it was thought that it would make sense to finish its development and bring it to market.
Frank Wyle, the President of Wyle Labs, a bit reluctantly, approved funding for completion of the calculator to the tune of almost $1 million, which, in 1962, was a great deal of money. Jack Rosenburg was brought in as the Chief Engineer to oversee the engineering aspects of the acquired company, including the calculator project. During Rosenburg's reign, he had brought in an engineer named Carlos Tomaszewski. Prior to working for Rosenburg, Tomaszewski was working at Wyle Labs as the Chief Engineer of Wyle's Products Division, developing data acquisition and telecommunications products. Tomaszewski's role under Rosenburg was to spearhead the engineering work needed for completion of the calculator project. Not that long into the projects, disagreements between Rosenberg and Frank Wyle left Rosenburg frustrated, and he left the company. With Rosenburg's departure, Tomaszewski assumed Rosenburg's role as Chief Engineer. Elmer R. Easton, who had been courted by Wyle Laboratories for some time, joined the company in 1957 as a marketing and general management executive. Prior to Wyle Labs, he had worked for Lear, Inc, selling Lear jets to corporate and political big-wigs for five years. During his time with Wyle Labs, Easton worked on a number of high-profile projects for the company, including work at NASA's Huntsville, Alabama Rocket Engine Test Facility, where Easton worked with the brilliant rocket engineer, Werner Von Braun. When Easton learned of the calculator project at Wyle Labs, he took a keen interest in the project, and ended spending some of his time involved with the team working on the calculator, providing managerial and project management guidance along with providing direct linkage from the calculator team to Wyle Labs' senior management advising them on the importance of the calculator project. Easton had a great deal of technical expertise from his former work at Curtiss-Wright Aircraft's Electronics Division as an engineer developing analog computers for flight simulators and jet engine controls. He had a keen engineering mind, along with a strong business acumen, and quickly integrated in with the calculator team. With a great deal of hard work from everyone involved, and great representation via Easton to the executive management of Wyle Labs, the calculator project was completed, with announcement in April of 1964. The calculator was designated the Wyle Laboratories Scientific model WS-01.
The WS-01 was quite an amazing machine for its time, utilizing an unusual design of a rotating magnetic disk (remotely related to the hard-disk drives found in a personal computer today) with seven dedicated tracks (with each track having its own read/write head) for storage of the working registers of the machine. A dedicated eighth track contained a pre-recorded timing track (recorded at the factory when the disk storage unit was manufactured) that served as the master clock for the logic of the calculator. This track also has a read/write head, but the circuitry for the write driver for the head was not included in the calculator circuitry to avoid any chance that the timing track could be written on. If the timing track were to be written on or damaged, this would render the calculator useless, as without its master clock signal, none of the rest of the circuitry could operate properly.
The logic of the WS-01 was implemented using discrete Germanium transistors. The calculator used a CRT display that provided on-screen display of the six calculator registers, each with a capacity of 24 digits. Unfortunately, there were reliability problems with the magnetic disk storage device, and a redesign was done to replace the troublesome disk with a type of storage that was more stable.
By late 1964, the Wyle Scientific Model WS-02 was introduced. The primary design change was replacement of the magnetic disk memory with a more reliable magnetostrictive delay line for the working register storage. Magnetostrictive delay line technology uses transducers positioned at the ends as well as taps along the way, of a long coiled piece of specially formulated metal wire. One transducer takes an electrical pulse and turns it into a momentary twist of the wire, creating a torsional pulse that travels through the wire at a specific speed (delaying the pulse), and is picked up by one or more similar transducer(s) after a period of time (or delay). The receiver transducer(s) generate a small electrical current when the pulse passes by, which is amplified and sent back into the electronics. The wire is of sufficient length that pulses representing all of the digits of the working registers in the calculator can fit within it at any given point in time.
The WS-02 was identical physically and functionally to the WS-01, and had exactly the same features. Electronically, the main difference was the change to the logic to accommodate the delay line memory versus the disk memory system, as well as additional circuitry to generate the master clock signal for the calculator, which came from a dedicated timing track on the magnetic disk of the WS-01 calculator and needed to be generated differently in the WS-02.
For both the WS-01 and the WS-02 Scientific calculators, the display showed the content of the three memory registers, along with the entry register, the accumulator, and the multiplier/quotient[MQ] register. The machines provided the standard four arithmetic functions, plus single-entry squaring and automatic square root. Both calculators were designed to allow add-on peripheral devices such as printers, external displays, a punched card reader, and paper-tape reader/punch equipment. When equipped with the optional external optical punched card reader, the calculators became programmable via standard IBM-style pre-scored punched cards that could hold up to 40 program steps per card. Cards could be taped together to allow arbitrarily (within practical limits) long programs to be created. Cards could be taped together in a loop, which allowed repetitive operations to be performed automatically. The card reader could read both forward and backward, allowing true branching and looping operations to be performed.
At the time that WS-02 was introduced, there weren't very many players in the electronic calculator biz. Sumlock/Anita in England had their Anita C/VIII, Friden with their EC-130, Mathatronics with the Mathatron calculator, and Wang, with their LOCI-2 machine in the US; and Casio and Sharp in Japan with their early transistorized calculators. Only Wang's LOCI-2 and the Mathatron were programmable in any way. Wang's LOCI-2 was big, expensive, and somewhat difficult to operate, even though it was technically more capable than the Wyle machine. The Mathatron was a technological tour-de-force, leapfrogging the state of the art, but the company that made it was a relative newcomer to the market, without a lot of exposure at the time.
The WS-02 was marketed and sold until sometime in late '67 to early '68, by which time electronic calculator technology had evolved to early '68, by which time electronic calculator technology had evolved radically, making it more and more difficult to sell the machine amidst the competition. A project to develop a new calculator was started, but never really made it off the ground, as Frank Wyle was reluctant to invest more in the calculator business.
The inability to convince management that a new calculator project was needed to replace the dated WS-01/WS-02 machines didn't mark the end of electronic calculators at Wyle Labs, though. Sometime in late 1965, Wyle Laboratories was contacted by Nippon Calculating Machine Co. (NCM), of Tokyo Japan. NCM had been very successful in the Japanese mechanical calculating machine market, and in July of '66, entered the electronic calculator marketplace with the introduction of its own desktop electronic calculator, the NCM Busicom 161. This calculator utilized transistorized circuitry and a small (256 bit) magnetic core memory for working register storage. Interestingly, the Busicom 161 shares a great deal of basic design methodology with the early calculators made by Industria Macchine Elettroniche (IME) of Italy. There were accusations by IME that the Busicom 161 was a cheaped-down "clone" of its elegant and extremely well-built IME 24, but nothing ever really came of it. Whether or not this was the case, the story appears to be lost in time. The reality was that the Busicom 161 was significantly less-expensive than other electronic calculators of the time, and was very successful. However, NCM knew that it couldn't rest on its laurels, as the electronic calculator market was expanding rapidly. There was a realization that the market lifetime the 161, and a lower-cost introduction with less capacity, the Busicom 141, was somewhat limited. NCM's management wanted a more advanced electronic calculator, and realized that it may not have all the expertise it needed to develop a more advanced machine. A search was begun by NCM to seek someone that could design and develop an advanced calculator for them.
Wyle Laboratories had embarked on making more of a market for itself by selling OEM (Original Equipment Manufacturer) products based on the Wyle WS-01 and WS-02 calculators. The design of the Wyle calculators was quite modular, and so Wyle offered components of the calculators to OEM customers. For example, the "calculating engine" of the WS-02 was offered as a stand-alone unit called the AP-01. This system could accept commands as if entered from a keyboard, and also had a coded output representation that could be used to get results of calculations. All an OEM customer had to do was add power, and interfaces to external equipment, and they had the computing capability that could be applied to any number of different applications including process control, data acquisition, and many others. It was this aspect of Wyle Labs business that attracted the attention of NCM.
The prospect of doing calculator development for other companies was something that Frank Wyle and the management team at Wyle Labs could get behind. Such development work would keep the Wyle Labs calculator engineers employed and busy, yet didn't involve all of the overhead of maintaining a calculator sales, support, and service business. There were thoughts that perhaps this role could be extended to other calculator companies, potentially bringing in some well-needed cash, while reducing overhead. Busicom representatives met with Wyle Labs, and soon an agreement was forged by which Wyle Labs would develop a next-generation calculator for Busicom, which Busicom would manufacture, market, and support as its own advanced calculator.
There was great relief among Wyle Labs products division employees when this deal was announced, and though there were some reductions in staff, this development did allow many of the core engineering team to remain employed. Work on the project began immediately. Wyle's engineers dusted off some of the concepts from the canceled follow-on to the WS-02, along with making use of new integrated circuit technology to create the design of the for NCM. The calculator design utilized a magnetostrictive delay line similar to that used in the WS-02. The use of new DTL (Diode-Transistor Logic) small-scale integrated circuits meant that the calculator was smaller and faster, as well as more-reliable than the earlier transistorized machines. This calculator design was licensed to NCM/Busicom for manufacture and sale by exclusively by NCM. The calculator was a five-function machine, with square root and two memory registers, utilizing a CRT display with vector-generated segmented digits (versus the beautiful sine/cosine generated digits of the WS-01/WS-02 calculators). The machine was introduced by NCM in November of 1967 as the Busicom 202, and sold solely through NCM's network of distributors.
While the Busicom calculator project kept Wyle Labs involved in calculator technology, during the development of the project for NCM, Wyle management kept steady pressure to move away from the calculator and in-house developed data processing equipment business, deciding instead to focus its Products Division on computer-based data collection and instrumentation systems, utilizing various commercially-marketed mini-computers versus its internally-developed data processing devices. Even though the numbers brought in by the calculator part of the Products Division were better, they still were not good enough, and it soon became clear that the days of calculator development work at Wyle Labs were running out.
The 202 was a good start on the advanced calculator that NCM wanted. However, Busicom wanted to advance its line of calculators with ever-more sophisticated machines. Wyle Labs was signed up in mid-'67 to develop machines with additional capabilities. The resulting machines were outgrowths of the technology used in the 202, and became the Busicom 207 and 2017 calculators. These machines provided additional memory registers for use in more complex calculations. The Busicom 207 came with with seven memory registers, and the Busicom 2017 with seventeen memory registers, both introduced in February of 1969. The 207 and 2017 calculators used a CRT display similar to that of the 202, and were similarly programmable via a built-in punched card reader.
As the calculator development projects for NCM wound down in mid-1968, Wyle Labs management decided to abandon any further business involvement in the calculator field. At the time, the last of the NCM project was transitioning out of Wyle to NCM, and it seemed a good time to make a clean break from the calculator business. NCM/Busicom was notified that there would be no more calculator development work done for them, or anyone else, and the agreement between the two companies was nullified. As it turned out, it wasn't such a big deal for Busicom, because it was already working on forging a relationship with a fledgeling Large Scale Integration IC manufacturer in the US.
This final decision by Wyle management to get out of the calculator business completely was of deep concern to a group of employees involved in Wyle's calculator business, who felt that the calculator market still held the potential for a successful business. By mid-'68, this group of folks decided they wished to venture out on their own to form a company that could focus on the electronic calculator marketplace. Foremost among these people was Elmer Easton, who was the prime instigator of the move to form a new company, as well having some financial resources that could potentially get the new business rolling. Easton and a core group of business-minded folks from Wyle Labs decided to put together a proposal for Wyle Labs management to spin off a new independent company to design advanced electronic calculators using the latest technology.
The proposal was drafted and presented to Wyle Labs management sometime in the late summer of 1968. Frank Wyle, the owner and President of Wyle Laboratories, blessed the proposal, and Wyle Labs became a major initial investor in the new company. Mr. Easton was named the President of the company. On the engineering side, Carlos Tomaszewski also joined the effort, along with a group of other talented businessmen, engineers, technicians and support staff.
The group of folks that left Wyle Labs formed "Computer Design Corporation", (hereafter referred to as "CDC", with apologies to the former Control Data Corporation [RIP]). Computer Design Corporation was incorporated in September of 1968, and initially located in Santa Monica, California. Straight away, CDC did some top-secret computing-related development work for the US government with early MOS(Metal Oxide Semiconductor) small and medium-scale integrated circuits. The government was very interested in integrated circuit technology for a wide range of military and intelligence applications, and CDC was provided some funding by the government to develop electronic computing technology related to cryptography. This work proceeded at a rapid pace, since many of the concepts had already been developed as part of the calculator work that was done at Wyle Labs.While there was great technical expertise at CDC, the company did not have a sales organization. However, it really didn't need sales and marketing, as the plan was to market the calculators the company would manufacture to companies that already had established marketing, sales, and service organizations. Given that there was already quite a bit of competition in the electronic calculator market, the founders decided that rather than selling calculators directly, they would sell their leading-edge technology calculators to other calculator companies, especially those that might be mired in the difficult transition from the mechanical to the electronic age of calculating. The development work done for the government fit into this model very well. The design developed for the government work ended up being a microcoded general purpose calculating engine, utilizing 64-bit floating point math, and recirculating shift registers for data storage. The microcoded nature of the calculating engine made it possible for it to perform a very wide range of calculating tasks. The next step, since the technology was already there, was to find customers who would be interested in selling and servicing the machines as OEM customers of Computer Design Corp. A prototype calculator was quickly put together utilizing small and medium-scale MOS integrated circuits, based on the designs developed for the government. This prototype was carried around in a large trunk to various potential customers to demonstrate the capabilities of CDC's technology. A road trip was made beginning in the early winter of 1968, with stops at many different calculator manufacturers, with the intent being to convince them of the market viability of Compucorp's advanced calculator architecture. Almost immediately, two potential customers expressed a great deal of interest: Deitzgen, an old slide-rule and drafting equipment company based in Chicago, IL, and, interestingly, former Wyle Labs customer, Nippon Calculating Machine Co, in Japan.
Another potential customer was shown CDC's calculator architecture in early 1969. This customer was Monroe Calculating Machine Co. Monroe was a very successful calculating machine company, but that success was based on mechanical and electro-mechanical adding machines, calculators, and accounting equipment. By late '69, it was clear that electronics were quickly taking over for mechanical devices, and as a result, Monroe needed to quickly come up with a way to update their technology. Monroe's ownership by Litton Industries, a major aerospace contractor, resulted in the design of a couple of interesting early programmable electronic calculators called the EPIC 2000 and EPIC 3000, but these machines were large (requiring a suitcase-sized electronics package, connected to a sizable desktop electro-mechanical keyboard/printer unit), expensive, and utilized all transistor technology, which was quickly being displaced by integrated circuits. Monroe was very interested in forming a relationship with a company that was concentrating on utilizing the very latest in electronics technology to reduce the size and cost as well as increase the performance, of high-end desktop calculating machines. Computer Design Corp. formed an OEM agreement with Monroe -- a major coup in the industry for CDC.
While the customer relationship building was going on, work was rapidly progressing at CDC toward taking the technology of the prototype machine and integrating it at a higher level, utilizing the latest "Large Scale" integration methods to compress the calculating engine design down to a much smaller number of integrated circuit devices. As is usual, there were some design changes between the original prototype and what was to go into production. The prototype machine was developed from original logic diagrams made by the digital designers, and the "bugs" were worked out of the design as the prototype was built and troubleshooting was performed. Such a methodology isn't practical when large scale integrated circuits are concerned. Once a batch of ICs is made, there's no way to change the logic in the chips without making a whole new batch of chips -- a very expensive proposition. The solution was to develop a computer-based simulation of the chips, such that their function could be tested, and logic problems discovered before the chips were actually fabricated. The simulator was written in FORTRAN on a Control Data 6400 computer located at a Control Data Computing Center located nearby Computer Design Corporation's headquarters. The simulator could be used for two purposes, one to test the logic, and the other, to test the microcode that had to be developed as the "program" for chip set. The simulator made it possible to assure that the logic to be committed to the chips had the best chance of resulting in bug-free chips with as few re-runs as possible. Once the logic design was completed and tested, the next step was to translate the logic into chip designs for a set of chips that would implement the logic. This process was beyond the means of Computer Design Corporation. The chip set implementation was contracted to AMI (American Micro-systems Inc.), of Cupertino, California. This first chip set design (called the "HTL" chip set) was completed late in 1969. The chip set consisted of nineteen different chips (part numbered HTL00 through HTL18) that could be combined in differing ways to provide the functionality needed for a wide range of calculating applications. The HTL chip set was an example of early large scale integration, but in reality, the integration levels were not all that high compared to what we are used to today, with only 300 or so gates per chip. These limitations were dictated by the integrated circuit process yields at the time, reflecting the true "state of the art" in the late 1960's for production large-scale integrated circuits. The chips were made using PMOS technology, using dynamic logic with a two-phase clocking scheme. Initially, AMI was the sole-source for the chips. AMI also fabricated chips for military applications, and as part of the military's requirements, a second-source for the military chips was required. Because of this, a separate IC fabrication company was spun out of AMI called Garrett Microsystems. Computer Design Corp. management believed that having a second-source for the ROMs that were critical parts of their calculators was a good idea, and an agreement was forged with Garrett to produce the ROM chips using the AMI chip designs. As a result ROM chips with Garrett markings can sometimes be found in Compucorp- and Monroe-badged calculators. Sometime in late 1971, General Instrument (an early leader in MOS integrated circuits) became another source for some of the HTL chips, specifically, the serially-accessed static RAM chips used as the main memory in the calculators. GI-labeled RAM chips can be found in some of the later-model HTL-chip set-based CDC calculators.
During the time the chip set was being developed, CDC's engineers were also working at a feverish pace to develop the necessary firmware that would run on the calculating engine to allow it to perform a wide range of calculating applications. The firmware engineers developed and tested the calculator firmware using the simulation of the calculator's logic running on a medium-sized computer system. This way, the development of the firmware could proceed in advance of the actual chip set being available. Also, the operational aspects of the firmware could be tried and true before being committed to the mask-programmed integrated circuit ROMs.
Once the production of the HTL chip set had begun, two scientific calculator implementations based on the chip set had been developed. One calculator utilized Burroughs-made Nixie tubes for the display, and the other, a built-in drum impact printer made by Seiko. Both calculators were not user-programmable, as the chips for the LEMP (Learn Mode Programmer) that provided the programmability for the calculators were still in development. However, both the display and printing calculators were quite powerful, with a range of scientific and engineering math functions that were superior to most of the existing electronic calculators on the market. The printing calculator utilized a specific HTL-series chip to help with managing the printer, taking load off of the main processing chips for the real-time response required by the printer, and the display calculator used another HTL chip that managed the refresh of the Nixie tube displays, again reducing the load on main processing chips since the Nixie tubes had to be refreshed fast-enough so as not to appear to flicker to the operator..which meant that they had to be refreshed at a minimum speed of 30 times per second.
Production prototypes of the two machines were shown to Monroe management in December 1969, and were received with great excitement. Monroe was so excited by the machines that the agreement with CDC was agreed upon very quickly. Not only was Monroe excited at the prospect of being able to sell advanced calculators into a market that was clamoring for more competition to reduce prices for high-end calculators, but CDC was overjoyed, as Monroe was a part of Litton Industries, which was a large and very wealthy company, so if the calculators were successful, there could be a lot of money in the deal for CDC. Things happened so quickly that CDC really had to scramble to get production versions of the agreed-upon line of calculators ready to go, as there was still quite a bit of work to be done.
Full-scale production of a number of different models of calculators tailored to specific disciplines (Statistics, Business, Scientific) began as soon as the various firmware additions were made for the application-specific calculators. The design of the firmware was modular. The main functionality of the a basic calculator was coded into a number of ROM chips. This core functionality includes power up self-test and self-configuration setup; management of the display and printer; keyboard management; a pseudo-code interpreter that is used for executing user-written programs via the LEMP chips; function dispatch routines; memory management; a full floating-point Binary-Coded Decimal math package; a large number of utility routines for calculator operation; and linkage routines that would allow ROMs outside of thee core ROM-set to call routines inside the core ROMs. This core set of ROMs would be installed in all HTL-based calculators, providing the basic functions required to make a basic calculator fully operational. This ROM-set basically contained the basic "operating system" for a calculator. To add on additional functionality to the base calculator, additional ROMs containing microcode for the specific functionality, heavily leveraging the routines within the core ROM-set, were added. For example, a high-end scientific calculator might require full trigonometric functions, logarithmic functions, functions to convert degrees/minutes/seconds to decimal degrees and back, as well as functions for converting decimal degrees to radians, and back. All of this functionality would be coded up in a separate ROM or set of ROMs (depending on how big the code was) that provided all of these functions, using the common math and memory management functions of the core ROM-set. When the calculator was powered up, part of the power-up initialization code in the core ROM-set would search through the memory space looking for "add-on" ROM modules, and would incorporate their features into the calculator's functionality. Each of the "add-on" ROMs had identification data included within it that would allow the core ROM-set to identify it, and incorporate the functions it provided into the calculator's capabilities.
The drum and hammer type impact printers were supplied by Seiko Instruments in Japan, under an OEM agreement. Initial versions of the printers were noisier than desired, requiring some minor design changes to be implemented that used softer materials for the print drum drive gears.
The success of the marketing of the calculators by Monroe and Deitzgen had CDC doing very well, providing funding to grow the company in rather dramatic fashion. By mid-1970, it had a quite number of customers that were successfully selling the company's advanced calculators, including Monroe, SCM, Deitzgen, Seiko, Ricoh, and Sumlock.
Monroe marketed a large line of the Compucorp-made calculators, including both Nixie-display and printing models. These machines were marketed as the 1600 and 1700-series calculators. Included in the series were the 1610 Scientific, 1650 Scientific, 1651 Scientific, 1652 Scientific, 1655 Programmable Scientific, 1656 Programmable Scientific, 1660 Printing Scientific, 1665 Printing Programmable Scientific, 1666 Printing Programmable Scientific, 1710 Scientific, 1765 Surveyor, 1766 Surveyor, 1770 Scientific, 1775 Programmable Scientific, and the 1785 Printing Programmable Scientific.
The success of Computer Design Corp. attracted the attention of the parent company of Monroe, Litton Industries. Merger discussions began, but the two parties could not agree on terms of a merger. The discussions ended with no deal. At this point, Computer Design Corp. management decided that it had sufficient resources to market its calculators under their own brand. In mid-1971, a division called Compucorp was created, and CDC began marketing and selling their own calculators under the Compucorp brand name.
The first series of calculators sold under the Compucorp name was called the 100-series. The line included a wide range of machines, all based on the same basic design using the HTL chip set, with the major functionality of each machine defined by the microcode burned into ROMs. There were two different form-factor of calculators in the 100-series, with the display type occupying less space, utilizing a Nixie tube display for presenting the user with the input to and answers from calculators. The 100-series also offered a series of larger footprint calculators that utilized the now-famous Seiko/EPSON drum and hammer-style impact line printer to provide printed record of the user's calculations (and programs). The 100-series display-type calculators consisted of the model 110, 110G, 110/X, 120, 121, 122, 122E, 125 and 125E Scientist calculators for scientific and mathematics applications. Statistical applications were supported by the Statistician series of machines, including the model 100, 130, 131, 132, 135, and Model 140 display calculators; and the model 141, 142, 145 and 145E printing calculators. Rounding out the line was a printing-only machine for specifically designed for surveying calculations, which was the Model 155 Surveyor)
Some time after the 100-series had been on the market, the 200-series was introduced. The 200-series calculators still used the HTL chip set, but new larger capacity microcode ROMs and read-write memory (RAM) chips allowed the machines to address markets that weren't able to be addressed by the 100-series calculators.
The 200-series calculators use the same mechanical construction, and all of the circuit boards were the same other than the memory (ROM/RAM) boards. These calculators focused mainly on getting Compucorp calculators into the financial applications realm. The 200-series machines retained the desktop, AC-powered form-factor of the 100-series machines, and continued to use either Nixie tube numeric displays, or Seiki/EPSON drum-type impact printers. A wide range of models made up the 200-series, including the 250 and 250-1 Bond Trader (used in financial market trading applications) machines; the 260, 261, and 263 Accountant calculators; the model 271 and 275 Banker calculators; the model 280, 281, and 285 Financier (with loan/mortgage type functions); and the model 291, and 295 Treasurer machines. Most all of the programmable 200-Series machines were offered with extended memory, which doubled the amount of program/memory storage space available. This upgrade involved simply plugging in an different memory board. The models which came from the factory with this extended memory added an "E" at the end of the model number. For example, a Model 275 Banker with extended memory would be designated a Model 275E Banker.
During 1972, while Compucorp's and its OEM customers were busily selling the 100 & 200-series machines, engineering was working on the next generation of Compucorp calculators. A new chipset was designed, taking advantage of increased levels of integration avaialble through Compucorp's primary chip fabricator, AMI. The new design ended up reducing the number of chips required to build a high-end programmable calculator from around 37 large-scale chips to as few as 20, including ROM and RAM. This new chipset would provide a more powerful instruction set, as well as a larger memory address space to support larger end-user programs in the calculators built with the chipset. Tom Scuitto, who had left Wyle Laboratories in 1966 (after the development work on the Wyle Labs WS-01 and WS-02 calculators had been completed) to sail around the world with his family, had returned from his trip, and provided consulting to Compucorp to help with the development of the new chipset. This chipset was called the "ACL" chipset. The ACL chipset was used in the 2nd generation of Compucorp and OEM partners' calculators. The ACL chipset was initially fabricated by AMI, but later, Texas Instruments was added as a second source to AMI, both for additional production capacity, as well as to safeguard Compucorp from any problem with AMI being a single source for their bread and butter, as well as creating some competition to help keep the IC costs under control. The chips made by TI had "TCL" part numbers. One interesting thing to note is that the Texas Instruments-made chips have different pinouts than the AMI-made "ACL" chips, though logically they are identical to the AMI chips. The pinout of the chip was determined by the chip design, and the AMI and TI designs were different enough that the pinouts varied between the two manufacturers. Compucorp's circuit boards were designed so that either maker's chip could be used, with different mounting locations for the TI- and AMI-made chips. Along with the pinout differences, the Texas Instruments-made chips also drew less power than their AMI-made counterparts, making the TI chips a preference for use in portable, battery-powered calclator applications.
The ACL chipset consisted of eight different chips, designated ACL/TCL 01 through ACL/TCL 08. The ACL 01 chip was primarily involved with driving the Seiko Seikosha (EPSON) line printer used in the printing models of calculators. At this point, it isn't clear if Texas Instruments produced a version of the ACL 01 chip. The ACL/TCL 08 chip was used in display-type calculators to drive a multiplexed seven-segment display panel, and was not used in the printing calculators. The ACL/TCL 02 chip performed the function of scanning the keyboard of the calculator, debouncing the keyboard switch closures, and translating keypresses into keycodes representing the keypress, as well as providing an input interface for external devices such as a punched card reader or magnetic cassette interface that could load or store programs into to the calculator's memory. The ACL/TCL 03 chip primarily functioned as the memory interface for the processing engine of the calculator. It would manage memory adddressing, reading microcode from microcode read-only memory(ROM), reading and writing data to/from the random access memory(RAM) as well as including some of the timing circuitry for generating memory refresh cycles when dynamic memory chips are used for RAM. The four-chip combination of ACL/TCL 04, 05, 06, 07 chips formed the processing engine of the chipset, including microcode decode and sequencing(04), index register and index adder(05), data routing and main adder(06), and control logic for the ACL/TCL 06 chip(07). These four chips combined to form the basis of what was essentially a complete 8-bit microcoded CPU, with performance and capabity that well exceeded that of Intel's 8008 single-chip eight-bit microprocessor.
There were two lines of 2nd generation Compucorp (and OEM's) calculators using the ACL/TCL chip set. The 300-series brought Compucorp calculators into the "handheld" realm. These machines would fit in a hand, but were a pretty serious handful. The term "portable" more aptly describes the form of the 300-series machines.
The 300-series machines provided similar functionality to the desktop 100-series calculators, but in two different form factors. One was the conventional desktop format, but with a newer, more modern-looking design, and keyboard keys that were of a completely different design than the earlier calculators. The other form of the 300-series calculators packed the functions into a much smaller, easily portable, battery-operated package. The advances in integrated circuit technology that were incorporated into the ACL chip set allowed a practical advanced battery-powered calculator to be made. Both form factors abandoned the expensive and somewhat fragile Nixie tube displays of the earlier models, replacing them with Burroughs Panaplex II planar gas-discharge seven-segment displays. The desktop calculators stuck with the same basic design of the previous-generation machines, utilizing either a Seiko/EPSON Seiko drum-type line printer, or for display calculators, the wonderful Nixie-tube displays used in the earlier generation calculators.
The portable machines included a battery compartment into which either four disposable D-cell batteries, or rechargeable Nickel-Cadmium cells could be installed. The Nickel-Cadmium batteries that were supplied with the portable calculators were manufactured by Gould, and had a sticker on them identifying them as Computer Design Corporation Part Number 3400017. An external power-pack provided ability to operate on AC power, as well as serving as a charger if Ni-Cad batteries were installed in the machine. These portable machines were hugely popular, especially with people who had to perform complex calculations "in the field", such as surveyors and construction workers.
The 300-series calculators were marketed both by Compucorp, as well as through Monroe, and Smith Corona/Marchant (SCM). The portable calculators in the series included the Model 320, 320G, 322, 322G, 324G, 326, and 326G Scientist calculators; the Model 340, 342, and 344 Statistician machines; the Model 360 and 360/65 Bond Trader calculators; and the Model 354 Surveyor. The desktop calculators included the 325 Scientist and the 327 Scientist. There are likely other models of the desktop machines that provided Statistical, Financial and possibly Surveying functions, but their existence is not entirely clear at this writing.
Some of the programmable 300-series machines provided the ability to connect a specialized cassette recorder in order to provide off-line storage for programs and data. This proved to be a real advantage when volatile RAM was used to store programs. Along with cassette tape program and data storage, there were numerous other interfaces for the 300-series calculators. Peripheral options included the Model 300 I/O Writer, the Model 310 Data Coupler Interface, and the Model 395 Teleprinter-Serial I/O Interface.
The 400-series of Compucorp calculators were introduced in the summer of 1972. These machines were similar in appearance to the earlier 200-series printing calculators, but utilized the new ACL chip set, but had more ROM and RAM capacity than the 200-series machines, but only desktop printing calculators were offered, forgoing the earlier Nixie tube display machines. These calculators were advanced and very capable devices, with greater I/O interfacing capabilities and significantly more program step and memory register storage. Total memory registers could be expanded in banks of 64 registers, up to a maximum of 522 memory registers (including the ten standard memory registers). These initial calculators in the line included the Model 425 Scientific, 445 Statistician, both with a starting price of $3,750. Shortly after introduction, the model 485 was introduced, providing financial calculating functions. Due to the expanded I/O capabilities of the calculators (mainly a result of the improved capabilities of the ACL chip set) Compucorp marketed the 400-series machines as computing systems, including a calculator "CPU" connected to peripherals such as floppy disc drives, line printers, and even CRT terminals. The 400-Series calculators were also marketed by Monroe as the 1800-Series, which included the Model 1810, 1830, 1860, and 1880 desktop calculators. The computer-like features of the 400-series calculators provided very flexible peripheral interfacing, allowing for wide range of peripheral devices to be connected to the calculator. All of the 400-series calculators came with a built-in 21-column, three line-per-second Seiko/EPSON drum impact printer, and a magnetic card reader/writer. The Compucorp 485 (also sold by Monroe as the Model 1830) was a general-purpose desktop calculator, with full programmability and I/O interfacing capabilities. The Model 425 Scientific calculator (sold by Monroe as the 1860) provided advanced scientific calculating capabilities, including built-in trigonometric functions. Lastly, the Model 445 (sold by Monroe as the Model 1880), was designed for advanced statistical applications. All of the Compucorp 400-series calculators could be combined with peripherals and specialized software to form fully-featured computing systems. The peripherals from the 300-series calculators would interface with the 400-series calculators, along with a line of 400-series specific peripherals that could push the calculator firmly into small computer system territory. These peripheral options included: Mark-Sense Card Reader (Model 490); Dual 8" floppy disc data storage systems (Model 491-2); Cassette Tape Unit (Model 492); X-Y Plotter (Model 493); IBM Selectric I/O Typewriter Interface (Model 495); and a 1/2-inch Magnetic Tape Drive (Model 497). Outfitted with the right selection of peripherals, a 400-series calculator could serve as the core of a capable business computing system.
During the course of 1972, the calculator market began to change quite dramatically. Single-chip calculator implementations were becoming commonplace from makers like Texas Instruments and Mostek. The prices for basic four-function calculators dropped dramatically over a very short period of time. Battery-powered, LED-display, handheld "four-banger" calculators became mainstream. The dramatic changes in the marketplace began to put pressure on makers of higher-end machines to reduce their prices. Then, like a bolt out of the blue, Hewlett Packard introduced its ground breaking HP-35, a rechargeable battery-powered, "shirt pocket-able", handheld scientific calculator. The introduction of the HP-35 marked the beginning of a calculator market shake-out, with many of the eager start-up calculator manufacturers falling by the wayside. Even some of the larger players in the calculator business started to suffer. Wang Laboratories, one of the major players in the early days of electronic calculator technology, dropped out of the calculator market by the mid 1970's as a result of the shake-out. The shake-out hit Compucorp (and it's partners) pretty hard. It became more and more difficult for the salespeople to sell a large, power-hungry desktop or "luggable" calculator that had a selling price in the $600 to $1200 range when anyone (who could stand to wait for delivery due to the rampant demand) could go out and buy an HP-35, which was much smaller, lighter, and nearly as capable (though lacking programmability), for a mere $395. Soon afterward, Texas Instruments introduced its own handheld scientific calculator, undercutting HP's price.
In an attempt to leverage its calculator business, yet move more toward the new market of small business computing, Compucorp marketed three bundled systems based on the 400-series calculator architecture. These bundles were called the Model 402 Business Computer System, the Model 403 Data Collection System, and the Model 450 Computing Calculator System. The 402 Business Computer System packaged a Model 485 calculator with the 491-2 dual floppy disc drive system, and a Digital Equipment (DEC) LA-36 120-column, 30cps dot-matrix printer, along with a software operating system called DOS-K that provided powerful file-oriented data handling capabilities for the very computer-like system.
The Model 403 Data Collection System was marketed as a multi-user data entry system to capture data entered by date entry clerks to magnetic tape for processing on mainframe computer systems. The 403 utilized a 400-series calculator as the CPU and provided connections for up to four CRT-display data entry terminals. The system came with the Model 497 1/2" Mag-tape drive, to which data entered on the terminals would be written for transfer to the larger computer. Also included was the Model 491-2 dual floppy disc drive system. The data entry forms definitions were stored on floppy discs to make it easy to change forms for differing data entry operations.
Lastly, the 450 Computing Calculator System was a multi-purpose system that could be configured for Business, Scientific, or Statistical applications. The 450 utilized a special CPU unit, based on the 400-series calculator architecture, but packaged as a monolithic unit that had no keyboard, printer, or magnetic card drive, and utilized much different firmware than the calculators. This monolithic CPU was called the 3000-series microprocessor. The CPU provided more RAM and ROM memory capability than the calculators could provide, allowing for more complex programs to run on the system. The systems came standard with 8K-bytes of user RAM, dual floppy disc drives, dual telecommunications controllers, keyboard console, magnetic card reader, and what was called a "System Printer", identical to the printer used the 400-series calculators. All of this was packaged into a desk-like form-factor. Along with the hardware, the DOS-K or DDD disk operating systems were available, along with an assembly-language processor called AL400 that could be used by programmers to develop complex applications for the system. Options for the system included a CRT-display unit, and a DEC LA-36 printer.
Even with the change of focus toward providing computing solutions, Compucorp's direct sales business began a slow decline through the early '70's, driven partly by competition from other calculator and small computer makers, and also due to a very aggressive sales office expansion around the world, which was quite capital intensive. By 1974, the losses incurred from the direct sales efforts forced Compucorp to seek a cure for the money drain. On Friday, August 2, 1974, Compucorp announced that it had reached an agreement with Litton Industries, Inc., the parent company of Monroe, that Monroe would become the exclusive distributor of Compucorp calculators in the US and Canada, with Compucorp agreeing to cease marketing calculators under the Compucorp brand name. In return for this rather drastic action, Monroe would place orders for $13,000,000 worth of calculators, and Litton would provide $1,400,000 in addition to $1,000,000 provided prior to the announcement of the agreement. Litton also agreed to purchase $1.125M worth of Compucorp stock, along with options for 250,000 additional shares exercisable at $4.50/share after one year. Most of Compucorp's sales and service employees were absorbed into Monroe. This action allowed Compucorp to focus on selling is business computer systems, while letting go of the overhead of marketing and selling calculators. The downside, which quickly became apparent, is that the real money maker for the company was the calculators. Calculators were easy to sell, and required little support once sold. Business computer systems were more difficult to sell, as the competition was relatively fierce, and frequently, special deals had to be made with potential customers that required customization and even custom programming, causing the sales cycle to be longer, and with that, revenues to take longer to be realized. It was also more expensive to sell the business computers, requiring more sales and sales engineering time for each sale. Along with this, the calculator engineering guys really didn't have all that much work to do, as the calculators that had been developed for Monroe were pretty much done, and since Compucorp was leaving the direct calculator market, nothing really new in terms of calculator development was in the offing. The bottom line of all of this is that the company began to steadily lose money at a rather spooky rate.
Because of the financial difficulties, Compucorp/Computer Design Corporation began to experience some of the dreaded brain drain, with some of the top calculator engineers leaving for companies such as Texas Instruments and Hewlett Packard. Compucorp's manufacturing arm kept busy building calculators for Monroe, but the losses mounted with only sales of its calculators outside of the US and Canada under the Compucorp brand allowed. Despite the difficulties, there was a core group of folks that were determined to keep the company alive, working intently to find a way to utilize the advanced microcoded chip sets for purposes other than calculators. If something could be found, it could potentially save Compucorp. While these efforts eventually yielded results, it came too late before drastic action had to be taken. Sometime in the latter part of 1975, after producing nearly 300,000 calculators (including those made for OEM customers), the Compcorp Division of Computer Design Corp. filed for Chapter 11 reorganization as a result of Security Pacific National Bank declaring a loan to the company in default. The terms of Compucorp's loan were that the bank would assume all assets of the company in the event of default, and this would mean that the company would effectively be liquidated. Liquidation simply was not an option as far as the core group of Compucorp executives and engineers were concerned.
The Chapter 11 filing prevented debtors from gutting the company, and allowed Compucorp's management time to re-invent itself. Prior to all of this difficulty, engineering had been experimenting with re-programming the ATL calculator chip set, which in effect was a multi-chip microprocessor, to become a rather elegant and effective word processor. The notion of word processing had begun to take hold in the business world, with early machines made by IBM (MT/ST, Magnetic Tape/Selectric Typewriter, 1964), Friden (Flexowriter/Justowriter, 1964), IBM (MC/ST, Magnetic Card/Selectric Typewriter, 1969), Wang Laboratories (1200, 1971) Lexitron (1972), Vydec(1973), and Linolex(1973).
With some tricky programming, and utilizing calculator interface circuitry that allowed a Diablo Data Systems (a company founded in 1969 a number of former Singer/Friden employees including George Comstock and Robert Ragen) HyType letter-quality daisy-wheel printer to be connected to the system, a rather capable word processing system was created, using essentially existing hardware. Computer Design Corporation had developed an interface that would connect to their HTL chipset calculators allowing the calculators to both accept and send data to an external device using RS-232 signalling and the 8-bit ASCII character set. The Diable HyType was available as a data terminal, with both keyboard and printer integrated into what looked like an ordinary electric typewrier, but was instead a high-speed (30 character per second) printing data terminal that could connect to computers and other devices using the RS-232 and ASCII character set. The HyType was a natural for connecting to the Compucorp/Monroe calcualtors with the calculator interface, and provided the ability for the calculators to generate extremely high-quality report type output, as well as even having the ability to generate graphics due to its fine carriage positioning combined with precise micro-stepping of the paper in both directions. Initially, the HyType targeted IBM's Selectic I/O devices, which were modified IBM Selectric II typewriters that had solenoids for activating the typeball's tilt and rotate positioning as well as providing carriage control and platen movement. The Selectric II provided the same flexibility of character sets with its unique easily-changeable "golfball" printhead and letter-quality output that the HyType provided, but the HyType was at least twice as fast as the Selectric I/O, operating at a minimum a 30 character-per-second printing rate versus the stated 15.5 character-per-second print speed of the Selectric I/O. The HyType also was significantly faster and more accurate at moving the paper, and could print with the carriage moving both the usual left-to-right, but also while moving the carriage right-to-left, providing true bi-directional printing capability, further adding to the effective printing speed depending on the type of output being printed. Compcorp became an OEM customer of Diablo Data Systems, providing the I/O Interface to the calculator, and the Diablo HyType data terminal as a package, which proved to be a very popular option for the calculators, since not only could the calculator print output on the terminal, but it could also accept alphanumeric input from the termina's keyboard, making it easy and convenient for providing a dialog-driven user-interface that was much easier for office staff to use than having to fiddle around with the calculator. Monroe was focused on calculators, and did not see any value in the notion of word processing, and thus was not something that Monroe would try to lay claim to. The prototypes were refined and turned into production word processor products. The Compucorp marketing and sales organization was rebuilt, specifically tooled to take on the competition and aggressively sell the new word processors. The calculator manufacturing lines were carefully retooled so that both calculators and word processors could be built on the same lines, without disrupting the flow of calculators for Monroe, but providing sufficient capacity to serve the fledgeling word processor business.
It was during this time, with Chapter 11 reorganization protecting the company from ruin, investment money from investors intriqued with the concept of word processing, began to come into the company to keep it afloat, although some stake was given up as a result. The Word Processor market was still in its infancy, but was a market that was poised to become huge. Compucorp was still in the calculator business, and because of contractual obligations, kept the calculator business in the forefront, while kfor the development of networking interfaces for Philips' products, as well as development of software that implemented an ISO-compliant network stack. The network stack became a product that was successfully marketed by Retix to a wide range of customers in the US, Europe, and Japan. Compucorp's success in the word processing market, offering word processing machines which were essentially running with the calculator chip-sets with new microprogramming that allowed them to operate as a word processor. The Compcorp word processing devices provided features that were not available from the competition (which included IBM and Wang Laboratories), the company began to see a resurgence in its financial situation, in time, returning to profitability. With Compucorp's engineering talent either leaving, or moving on with Retix, no further development effort in Compucorp's calculator technology was made. By 1984, with the advent of IBM's "PC", and the market for calculators reduced to a "short-list" of players such as Sharp, Casio, Canon, TI, and Hewlett Packard, the market for Compucorp's products was no longer viable, and Compucorp ceased operations.
There are many names of companies that participated in the calculator boom, and suffered in the later 'bust' of the calculator market...names like Bowmar, MITS, Cintra, Wang, Tektronix, Sony, and many more. Not all of these were complete victims, but certainly their forays into the calculator marketplace were folded up during the calculator wars of the mid-1970's. To this author, Compucorp was one of the more tragic losses of those times. Compucorp designers were truly ahead of their time, designing essentially the precursor to the microprocessor with their calculator 'core' chip-sets at a time when the notion of a microprocessor simply didn't exist. Unfortunately, market pressures and the rapid rate of change in calculator technology forced many technologically advanced players out of the market. Sadly, Compucorp ended up being a victim of the times.