Mathatronics Mathatron 8-48M Mod II Electronic Calculator
The Mathatronics Mathatron calculator is a historic machine, marking many firsts in the electronic calculator industry. While generally overlooked by history, the machines made by Mathatronics were way ahead of their time, offering features that in many cases were not available from other calculator makers for years. While this exhibit is related to the Mathatronics 8-48 Mod II, which was introdced a little over a year after the formal introduction of the original Mathatron 4-24 calculator, the exhibit attempts to capture as much as is known of the history of Mathatronics, and the development of their amazing calculators.
The Mathatron calculators underwent a great deal of change early in their lifetime, with the company introducing variations based on improvements in design and capability in rapid-fire succession. Some of the timeline is difficult to track due to a lack of definitive information, however, some history has been able to be recovered through discussions with one of the founders of the company, and also from vintage articles and advertising that has been found.
The first Mathatron calculator was the Model 4-24. This designation was due to the fact that the machine had four memory registers, and 24 steps of learn-mode program memory. This machine used two rotary switches on its front panel; one for the power switch, and the other to select the operating mode of the machine. The 4-24 was soon joined by a machine with expanded memory, designated the Model 8-48, which doubled the memory register and program step capcity. Both of these machines were four function machines, using full algebraic problem entry that followed the rules of arithmetic precedence, provided store/recall memory registers, and offered simple learn-mode programming which allowed the calculator to "learn" sequences of keypresses, then replay them to automate certain types of calculations.
The Model 8-48M Mod II is an improvement over the original design Mathatron calculators, the Model 4-24 and 8-48, by virtue of a minor redesign that made a few changes to the machine. The first change was that the original machines used rotary switches for the power switch and mode-selection function of the machine, and the Mod II machine changed these controls to pushbutton switches. The first generation 8-48M shared essentially the same "CPU" (main logic assembly) with the original Mathatron 8-48, with additional logic to provide the pre-programmed functions. The second generation 8-48M Version 2 utilized what was called the "Model II" logic unit, which incorporated some design changes to the logic unit to better integrate the pre-programmed functions into the operation of the machine, and handle pushbutton access to these functions rather than selection by a rotary switch.
Actual "ticker-tape" printout from Mathatronics Mathatron calculator
The pre-programmed functions of the 8-48M were stored in a magnetic core read only memory (ROM), holding normal Mathatron program steps which would carry out the various functions. Because the ROM could hold a selection of programs for any type of calculation, there were a number of different machines (both versions I and II) with ROM programs that were targeted towards various specialized math applications. The 8-48M was targeted toward mathematical applications. The model 8-48S had pre-programmed functions applicable to statistics work. The model 8-48C provided pre-programmed functions for civil engineering calculations, and lastly, the model 8-48SC provided functions useful for scientific calculations.
The Mathatronics Mathatron calculators hold a very important place in electronic calculator history. When introduced in late 1963, the Mathatron was the first solid-state, desktop, printing, floating point, algebraic entry, programmable, stored-program electronic calculator. That's quite a mouthful, but these dramatic (for the time) distinctions make the Mathatron the first production example of a desktop "personal computer" that could be programmed to automatically carry out complex calculations. Olivetti's Programma 101 is historically recognized as being the first desktop "personal computer", but the Mathatron was shipped to its first paying customer (Woods Hole Oceanographic Institute) nearly a year before the Programma 101 was announced. The Olivetti machine can boast an important first, that of providing for offline storage of programs and data via a magnetic card -- a feature that made it possible to quickly change the program in the machine without having to type it in each time. This first, combined with Olivetti's already solid position in the business machine marketplace, allowed Olivetti's machine to simply receive more attention than the Mathatron, which cemented its place historically, even though the Mathatron was truly the first of its kind.
The idea for the Mathatron was the brainchild of William Kahn, an electronics engineer who was frustrated by having to work with the slow and cumbersome electro-mechanical calculators of the day. Having worked on the design of first and second-generation computers at Datamatic (Honeywell's computer division), Kahn felt that many of the concepts of large computers could be scaled down to a small desktop-sized "personal" machine that would dramatically increase the productivity of people involved with performing complex mathematical operations as part of their work. Kahn worked in his spare time developing the ideas behind such a machine. After finishing up the design for Datamatic's H-800 computer, Kahn left Datamatic to go to work for Raytheon. Kahn had shared his ideas for a calculating machine with his former boss at Datamatic, Roy Reach, who seemed intriqued. At Raytheon, Kahn met David Shapiro, who also shared enthusiasm for Kahn's calculator concept. In February of 1962, the three men quit their jobs, and set out on their own to form a company with the goal being to make Kahn's calculator concept into a commercial reality. The men pooled their own money to come up with $18,000 of startup funding, and managed to round up an additional $54,000 in capital from outside investors. Roy Reach's wife, Marjorie, came up with the name for the calculator -- Mathatron. This served as the root of the name of the company, Mathatronics, Inc. The men set up shop in Waltham, Massachusetts, and used the initial funding to develop a working prototype of the calculator. The prototype would be used to demonstrate the concepts of the calculator to Mathatronics' outside investors, who were granted options in the company shortly after it was formed. These options would mature 30 days after the first successful demonstration of the prototype, and could provide the company the next infusion of capital to turn the prototype into a production product.
"Clamshell" design of the 8-48M Version 2, logic unit on bottom, keyboard/magnetic core/power supply on top
Kahn, Reach, and Shapiro worked tirelessly (and without pay) through late 1962 to develop the working prototype. In early December, 1962, their genius and hard work was rewarded in the form of an operating prototype calculator. This prototype machine had the basic form factor and features of the production calculator, but was housed in a wooden cabinet. The prototype provided four memory registers and 24 steps of "learn mode" program storage. The machine had decision-making features, allowing conditional and unconditional branching, making iterative calculations feasable. It provided printed output in the form of a fast 'ticker-tape' style serial impact printer. It had a capacity of nine significant digits, with full floating decimal, and a two digit decimal exponent ranging from -42 to +58. The machine featured algebraic entry, meaning that problems would be entered on the keyboad just as they would be written on paper, with the machine following the rules of precedence and parentheses. The calculator automatically performed the four standard math functions, along with an automatic single-key square root function. No other desktop electronic calculating machine at the time, or for almost a year afterward, could even come close to the capabilities of this prototype calculator.
The prototype machine, designated the Mathatronics Mathatron Model 4-24, was succesfully demonstrated to the investors a few weeks later, in late December of '62. The investors were amazed by the demonstration, and quickly offered to buy up their options in the company, immediately raising an additional $300,000 of capital to make the Mathatron a production reality.
In early 1963, Roy Reach started touring around the US with a non-functional wooden mockup of the machine, and examples of printed output from the prototype machine. While this was going on, Kahn and Shapiro were working on turning the prototype into a production reality. One of the places that Reach visited was Woods Hole Oceanographic Institute (WHOI). Based on the virtues of the paper tape output from the prototype machine, along with Reach's convincing pitch, WHOI placed orders for two Mathatron 8-48 (double the memory and program step capacity of the prototype machine), without ever seeing a real calculator.
At the time the order was placed, a lot of details had yet to be worked out for the actual production calculator. WHOI wasn't daunted by having to wait for a production machine, as the calculator was exactly what they needed for ship-board calculations for their oceanographic research. At the time, there simply wasn't anything else that could come close to the capabilities of the Mathatron that would fit within the confines of a sea-going research vessel.
Mathatron 8-48M Version 2 "Top Plate Assembly"
Left: Keyboard Encoder
Right: Core Memory Arrays
In June of 1963, Friden had publically demonstrated their groundbreaking solid-state calculator, the Friden 130. The Model 130 is historically recognized as the first desktop solid-state electronic calculator. While a major accomplishment, the features of the Friden 130 paled in comparison to the capabilities of the Mathatronics machines. And, at around the same time of the Friden 130's debut, Mathatronics had the first production Mathatron 8-48 ready to deliver to WHOI. Kahn and another Mathatronics engineer named Charles French, loaded the machine into Kahn's car, and set out for the roughly two hour drive to Woods Hole. Unfortunately, when the machine was powered up to show to the eager scientists at WHOI, it would't work. Some quick attempts were made at troubleshooting and repair, but these were unsuccessful. With their "tails between their legs", Kahn and French loaded the machine back into the car, and headed back to Waltham, determined to figure out what had gone wrong. A couple of weeks later, the problems had been identified and fixed. In early July of 1963, Kahn and French returned to WHOI with a fully operating machine. The folks at WHOI were enormously satisfied with the machine, and arranged for it to be installed onboard one of their research vessels for immediate use. Shortly thereafter, the second production Mathatron 8-48 calculator was delivered to WHOI, where it was immediately sent with meet up with the same ship somewhere overseas.
Keyboard implementation detail: Keycap actuated micro-switches
Technically, it could be argued that the Mathatronics Mathatron calculator should take the historical credit from Friden as being the first solid-state desktop electronic calculator, but the Mathatron was not publically introduced until November, '63 while Friden's EC-130 was introduced in June. Both Friden and Mathatronics (and probably some others) had prototype calculators running in late '62, but from a historical perspective, the formal public introduction of a calculator for sale seems to be the appropriate benchmark to use for determining who was first to market with a new technology.
At the same time that Mathatronics was working to get the WHOI order fulfilled, manufacturing capacity had ramped up, but the marketing effort had stalled somewhat. There were a few orders that had come in from other customers (MIT's Lincoln Lab and MITRE Corporation) as a result of Reach's low-buck marketing visits, but no concerted public marketing effort had yet been made. A decision was made to formally introduce and kick off the marketing campaign for the Mathatron at the NEREM trade show to be held in nearby Boston in November of '63.
Mathatronics "Model II" Logic Assembly
The NEREM show, along with other electronics trade shows where the Mathatron calculators were shown, were a big hit. By early 1964, orders began coming in at a good clip, and spirits were riding high. Mathatronics had made a place for itself at the high-end of the brand new electronic calculator marketplace.
Unique "bobbin wire-wrap" wiring method (US Patent 3,502,787)
Through 1964 and into 1965, the model 4-24 and 8-48 were good sellers. Mathatronics' engineering expertise went to work improving the original design, making minor changes to allow for the addition of external devices such as a teletype interace to allow a Teletype model ASR-33 to be connected to the calculator to provide better data printout capabilities, as well as development of a memory expansion unit (known as the Mathatronics APS) that would provide additional program step and memory register storage. Another improvement was the addition of a read-only memory (ROM) that stored 'canned' programs for more advanced math operations such as trigonometric functions, logarithms, and exponential calculations. These changes all came together to become the first generation Mathatron 8-48M(Mathematics), 8-48C(Civil Engineering), 8-48SC(Scientific), and the 8-48S(Statistical) machines. These machines had a rotary switch on the front panel that allowed the desired math function to be selected, along with a keyboard button that caused the selected function to be executed. These "Version I" machines and peripherals were introduced in the spring of 1964.
While the Version I Mathatrons were successful, the rotary switch based controls proved to be a bit tedious in day to day use. A minor redesign of the main logic unit allowed better integration of the advanced math functions. The "Model II Logic Assembly", along with an improved keyboard interface circuit board, provided the logic necessary to allow the rotary switch mode and function selection controls to be replaced by individual keyboard pushbuttons. This design change made the advanced math functions provided in ROM much more accessible to the user, with each function having its own key on the keyboard. With this change, a single keypress would trigger the execution of the ROM program for the desired function. The calculator exhibited here is an example of a Version II Mathatron calculator. The Version II machines were introduced in late 1964.
Ad for Barry Wright/Mathatronics' Mathatron 8-48M Version 2, Circa mid-1966
From an electronics design standpoint, these machines were based entirely on resistor-transistor logic(RTL). This type of logic was rather power hungry, and couldn't run as fast as the diode-transistor logic of machines like the Friden 130, but was the most inexpensive form of logic to use. In such a complicated machine, the savings of using a resistors in place of diodes meant a significant reduction in the overall manufacturing cost of the calculator. While resistors were much less expensive than diodes, there was some tradeoffs made in the design, as pure resistor-transistor logic requires more transistors, which were arguably the most expensive electronic component in the machine. There are over 1000 transistors in the main logic assembly of these machines. This doesn't count the additional transistors in the keyboard circuitry, power supply, and core memory (both RAM and ROM) in the machine. The transistors used in the main logic are primarily PNP Germanium-based alloy-junction 2N404 devices (introduced by RCA in 1957), one of the early transistors which became widely used for digital logic. By the time that these machines were in full production, the 2N404 had been around for quite a long time, and the price per transistor was likely somewhere in the $0.40 range when purchased in quantities. Transistors in the machines seem to come from a variety of manufacturers, including General Electric (GE), Radio Corporation of America (RCA), and General Instruments. From the Mathatronics factory, the tops of the transistors are painted white (potentially to obscure the part number to discourage unauthorized repair), and stamped (or in some cases, hand written) with a number to identify its function in the circuitry. The manufactures of the transistors have been identified from observing machines which have had transistors replaced over the years. Typically, the field-service folks who had to do this tedious component replacement work didn't bother repainting and renumbering the replacement transistor, making the replacement parts easily identifiable.
For a period of time, as yet undocumented, Mathatron calculators were built using hybrid circuit modules which contained multiple resistors in one monolithic in-line package. These packages would replace the vast majority of the vertically-mounted resistors in the main logic panels. These hybrid modules had solder pins on the bottom (toward the circuit board) and top, and the same wire-bobbin connections were used for connections to the top of the hybrids. Due to reliability issues, and the additional complexity of replacing an entire module when one component inside the module failed, the hybrid design was phased out of production after only a short periods, and discrete resistors again returned to use in the machines. The earliest Mathatrons used discrete resistors, then there was the period of time where the hybrids were used, followed by future Mathatrons again using discrete resistors. Mathatron calculators which use the hybrid modules are very rare today than those with discrete resistors. The hybrid modules came in a variety of standard packages with commonly used resistors in various arrangements to fit the bulk of the rows of resistors in the logic panels. There were still discrete resistors used in the hybrid-design machines, but far fewer. The sad part about any machine that has survived to today with the hybrid modules in place is that replacements for any failed hybrid modules are virtually impossible to find. The hybrid modules were custom-manufactured for Mathatronics, and because of the problems associated with the devices, they weren't used for long (probably less than 6 months of production), and spares were kept in limited supply. When a hybrid module failed in the field, it was typically removed and replaced with standard discrete component resistors.
This exhibit is a work in progress. More information has been gathered, and another Mathatronics 8-48S (Statistical) Version I has been acquired which will soon be added as an exhibit. Please check back from time to time for updates as they become available.