ICM (International Calculating Machines) Model 816 Desktop Calculator
This machine has some historical significance, as it was the first electronic calculator to be made using an "off the shelf" MOS Large-Scale Integration IC chipset. The machine was made by International Calculating Machines(ICM), of Woodland Hills, California, and was introduced sometime during the last half of 1970. The ICM 816 was the first electronic calculator marketed to use Electronic Arrays' new six-chip S-100 calculator chipset. ICM was a calculator manufacturing subsidiary of integrated circuit manufacturer Electronic Arrays(EA), located in Mountain View, CA. Of additional historical significance is the background behind the development of the chipset and the calculator has direct linkages back to the 1964 introduction of a very historical electronic calculator, the Victor 3900.
In the early 1960's, the management of Victor Comptometer, famous for its highly reliable mechanical and electro-mechanical adding machines and calculators, had become concerned that the days of mechanical calculating devices were limited. With the introduction of the Sumlock Comptomenter Anita electronic calculator, made with tiny gas-discharge thyratrons and semiconductor diodes, as well as rumors that a number of companies around the world were actively working on developing electronic calculators made with transistors, the handwriting was on the wall -- electronic calculators were the up-and-coming thing, and any company that made mechanical calculating machines had better get on the bandwagon, lest they be left behind.. Victor management instituted a crash project to develop an electronic calculator, but the prototype filled a small room, and used vacuum tubes, which were impractical for building a desktop calculator due to size, power requirement, and heat concerns. While it was a good test that Victor had the logic design skills needed to make an electronic calculator, they simply didn't have the expertise with transistor technology to develop a calculator using them, and it was unlikely that those with solid-state design skills could be hired quickly enough to meet the timelines required.
Enter a company called General Microelectronics (GMe). GMe had spun out of Fairchild (one of the early leaders in the Integrated Circuit market) with a prime interest in development of complex Metal Oxide Semiconductor (MOS) integrated circuits. Victor ended up contracting with GMe to develop a set of highly advanced MOS integrated circuits to create thier first electronic calculator, the aforementioned Victor 3900. The 3900 was not entirely successful, partly due to reliability issues (the IC's seriously stretched the state of the art in IC manufacturing), and partly due to the fact that the development of the ICs nearly bankrupted GMe, causing supply issues for the chips that made up the calculator. Despite its lack of real success, the Victor 3900 did set a benchmark in electronic calculator history -- it was the first electronic calculator implemented with integrated circuits, and to boot, they were MOS Large-Scale Integration devices. Sharp Corporation would introduce a calculator that they claim to hold this benchmark, the Sharp QT-8D, nearly five years later! In financial dire straits, GMe was sold to Philco Electronics in 1966, and at that point, a lot of the MOS IC design talent within GMe left.
Among those that left in GMe in 1967 was Dr. Jim McMullen, one of the founding members of GMe. McMullen brought with him a number of other GMe folks, and founded McMullen & Associates, a Large-Scale IC design company. The initial folks involved in the founding of Electronic Arrays were David Stiefbold, Earl Gregory, and E. Stephenson from GMe, as well as Sam Nissim and J. Nicklas from Bunker Ramo. After operating as McMullen & Associates for a short time, an influx of investment money from various wealthy sources transpired. As a result, the McMullen & Associates changed its name to Electronic Arrays(EA). With the investment money, the company was able to develop a state-of-the-art Large-Scale MOS IC fabrication facility which provided the cabability for Electronic Arrays to design and manufacture very complex MOS integrated circuits that stretched the state of the art.
At its beginning, EA developed MOS shift register ICs with large capacity, which proved very useful in computer and military data encryption equipment. Later, the company moved into development of high-capacity (1024 bits) Read-Only Memory (ROM) chips that could be used to store the operating code for computers. These early "larger scale" MOS IC developments got the company into a good revenue stream from the military industrial complex. That money provided a steady flow of funds into research and development, that led to continuous improvements in the company's MOS IC processes.
McMullen, who had been involved in the development of the Victor 3900 Large-Scale MOS chips at General Micro-electronics, had a keen interest trying to make a calculator chipset again, this time using the more refined and advanced MOS IC fabrication techniques that EA had developed. Beginning sometime in 1968, McMullen led a team of engineers who were put to the task of creating an electronic calculator chipset that would use as few chips as possible, at an extremely low cost for the time. To ease the design of the chips, a method similar to that used to develop the logic for the chips of the Victor 3900 was utilized.
Electronic Arrays' Calculator Development Platform (Note Pandicon display module)
A calculator development platform was built, based on off-the-shelf small and medium-scale integrated circuits placed on plug-in circuit boards. The calculator was purposefully designed to segment the architecture of the machine into individual functional units, such as input processing, aritemetic unit, control array, output generation, and register store, with the idea that each functional unit's logic could be used as a basis for design of a single chip with the functionality of that unit. Along with these boards was a device that simulated the Read-Only Memory(ROM) that the microcode for operating the calculator was contained within. This ROM simulator used Read-Write Memory (RAM) to subsitite for the ROM, so that modifications to the microcode could easily be made on the fly, with the changes observed immediately. The idea of the calculator development platform was that the functional units, along with the microcode that controlled them, could be thoroughly tested and debugged on the development platform. Once the design of a given functional unit was tried and true, the logic design of the functional unit could be translated into a Large Scale Integration MOS IC. This assured the design of each unit, along with the microcode that choreographed the operation of the functional units, was completely wrung out before committing the logic to the expensive process of developing prototype chips, maximizing the chance that the chips would work the first time. Once the microcode was thoroughly debugged, it was embedded into a Large Scale ROM chip that made up the sixth chip in the complete chipset. The use of functional units along with ROM-based microprogramming to control the operation of the functional units allowed great flexibility, as changes or improvements to a given functional unit could be designed at the large scale on the development calculator, proven out with extensive testing, and rendered into a new chip with the changes incorporated into it. Modifications would be made to the microcode to properly operate the new chip, and, along with the existing chips, a new functionality for a calculator could be created with relative ease. For example, a new output functional unit could be createad to drive a printer versus a display. This calculator development platform was used to create the initial design of the EA S-100 chipset that was produced for the ICM 816 calculator, as well as a number of improved and follow-on chips that added additional functionality, or, as higher levels of integration became available, allowed multiple functional units to be combined onto a single chip.
In November, 1969, Electronics Arrays announced general availablity of the EA S-100 six chip electronic calculator chipset. While Sharp had set the calcualtor market into a tizzy with the March, 1969 introduction (in Japan) of the Sharp QT-8D, which used an advanced 4-chip MOS chipset made by Rockwell, the Sharp chips were completely proprietary, while anyone could buy the EA S-100 chipset from Electronic Arrays to build their own calculator, a fact that EA made abundantly clear in their announcement of the chipset. The EA S-100 chipset was the first publically available calculator chipset. Prior calculator chipsets were proprietary devices that were for use strictly by calculator manufacturers in their own machines, and though many calculator makers had to rely on outside Large-Scale IC design and fabrication companies (For example, Canon used Texas Instruments) Electronic Arrays published detailed application notes for the S-100 chipset, written such that an electronics hobbyist could use the chips to build their own calculator. These applications notes were made available for a small price to anyone who wanted them. Electronic Arrays also offered custom chip design services to calculator companies so that modifications could be made at the logic and microcode level to customize chips at the request of a calculator manufacturer to allow special features to be included in their calculators. This led to variations of the S-100 chipset developed for calculator manufacturers to offer their own unique features not found on the generic S-100 chipset.
In the early part of 1970, Electronic Arrays created the International Calculating Machines, Inc. subsidiary to produce low cost calculators using the S-100 chipset. A facility in Woodland Hills, CA was established, and the process of putting a manufacturing operation in place was begun. By the fall of 1970, the factory had begun building the ICM-816 calculator. Along side the manufacturing operation, a sales and marketing organization was built up, and some time in the last half of 1970, the ICM 816 was introduced to the marketplace, as a price-leading, highly-featured desktop electronic calculator. The initial introduction price of $450 was rather steep, as Sharp was selling its QT-8D for $395. This steep introductory price proved to be rather a mistake, as buyers were much more likely to spend less money for a calculator made by a well-known manufacturer versus a relatively unknown company such as ICM.
The nameplate for the "Senator Mini Calc" version of the ICM-816
Serial Number Tag for Caltype Senator Mini-Calc
Sometime during the '71-'72 timeframe, fully assembled ICM 816 calculators, were sold under OEM contract to Caltype Corporation (a subsidiary of Transitron Electronic Corporation), of Los Angeles. Transitron Electronic Corporation was founded in 1952 to manufacture transistors, and did quite well for a time, but changes in transistor fabrication technology and the advent of integrated circuits eventualy led to the demise of the company in 1986. Caltype Corp. was a spinoff of Transitron, incorporated in September of 1965 as a marketer of low-cost business machines manufactured by other companies. Caltype marketed the ICM-816 as the "Senator Mini-Calc". The only difference between the ICM-816 and the Caltype Corp. Mini-Calc the name plate on the top cover of the machine, which reads "Senator Mini-Calc; Made in USA"; and the model/serial number tag, which replaced ICM's name with Caltype's. The model number of the machine listed on the model/serial number tag of the Caltype machine remains ICM-816, and the serial-numbering scheme follows ICM's sequence.
Also, sometime in the 1971 timeframe, another small company called Lago-Calc, Inc., showed up on the scene. It is interesting to note that Lago-Calc was located at 6109 De Soto, Avenue, in the same city as ICM, Woodland Hills, California. Lago-Calc marketed (among a relatively small line of calculators) a machine called the LC-816 that was functionally identical to the ICM-816, with slight differences in cabinetry and keyboard construction (both a bit higher-quality than the ICM machine). It appears that there were two different versions of the LC-816, one that used the origial Electronic Arrays S-100 six-chip set, and another that utilized an four-chip set also made by Electronic Arrays that provided the same functionality, but in fewer chips. The first version LC-816 utilized the same main circuit board and Electronic Arrays S-100 chipset as the ICM-816 and Senator Mini-Calc, but rather than using individual Nixie tubes for the display, the calculator uses an unusual display device called a Pandicon, which is essentially eight Nixie display elements packaged inside one glass envelope. As with the Senator Mini-Calc, the Lago-Calc LC-816 has a very similar model/serial number tag to the ICM-816, and the serial number scheme is also the same, with the first three digits of the serial number being 816. Sometime in 1972-1973, Lago-Calc was acquired by Facit-Addo, Inc, which was the US business machine subsidiary of Facit AB, itself a subsidiary of the massive conglomerate Electrolux. It appears that shortly after the acquisition, the company moved from its original location in Woodland Hills, California to a much larger facility in Canoga Park, CA, where both the machines originally made by Lago Calc, along with manufacture of mostly printing electronic calculators developed by Facit-Addo in Sweden. Along with the Lago-Calc LC-816, an Addo-X-badged version called the D1 was manufactured, identical to the LC-816, with only the model badge and serial number plate reflecting the Addo-X brand. Due to the poor economy, intense competition from Japanese calculator manufacturers, and the huge shake-out of the electronic calculator marketplace in the 1973-1975 timeframe, Electrolux closed down Lago-Calc sometime in 1975. If anyone out there knows anything about Lago-Calc, Inc., the museum would love to hear from you. Please click the EMail link at the top of this page to let us know what you know.
The S-100 chipset was also sold in quantity to a small company in California called MITS (Micro Instrumentation and Telemetry Systems), founded by Edward Roberts and Forrest Mims in December of 1969. By early 1971, the company was wholly owned by Roberts, who sold off 15% of the ownership of the company in order to raise funds to develop a calculator kit that could be assembled by anyone who was competent with a soldering iron. One of Roberts' friends had read about the Electronic Arrays chipset, and after a little research, it seemed the perfect basis for a MITS calculator kit. Roberts designed a calculator based on the EA chipset that could be built as a kit, or ordered as a fully assembled calculator. The MITS 816 was introduced in the November, 1971 Popular Electronics magazine, with an article going over the basics of the design of the calculator. The article provided information on how to order the MITS 816 calculator for $179 in kit form (with all parts included, and detailed assembly instructions), or $275 fully assembled and tested. The MITS 816 set a new low-price standard for a basic four function calculator. The MITS 816 used a vacuum fluorescent seven-segment display as opposed to the more expensive and somewhat more complex to interface) Nixie Tube display used in the ICM-816. MITS and Ed Roberts later became famous for developing the Altair 8800 microcomputer kit, that is generally considered the hobbyist computer that marked the beginning of the personal computer revolution.
It is also very interesting to note that world-famous electronics manufacturer Sony, (which entered the calculator marketplace in 1967 with it's ICC-400W and ICC-500W calculators), marketed a calculator called the ICC-88 beginning in fall of 1971 that utilized a variant of the Electronic Arrays S-100 chipset. The ICC-88 was Sony's first calculator to use non-Sony-made integrated circuits in one of their calculators. The display in the ICC-88 was a gas-discharge, seven segment planar display versus the Nixie-tube display of the ICM-816, and the Sony machine could operate on AC power, or through built-in rechargeable batteries. Otherwise, the ICM-816 and Sony's have the same features and operation.
Along with the Sony machine, Walther, the German company famous for making firearms, also used a variant of the S-100 chipset in their Nixie tube-based ETR-2 desktop calculator, introduced sometime in 1971, as well as the later (late '71 to early '72) Nixie-based ETR-3 rechargeable battery-powered calculator. Walther went on to produce quite a number of other calculators using variants or follow-on chipsets from Electronic Arrays.
As evidenced by the number of companies that leveraged Electronic Arrays' inexpensive calcualtor chipset, there was a dearth of upstart calculator manufacturers that were looking to try to grab a share of the market created by the dramatic decrease in price of electronic calculators made possible by large-scale integrated circuits. At the time, IC technology was advancing very quickly, which made it difficult for the smaller players to keep up with big boys in the integrated circuit business. Competition in the Large Scale Integration chip market was intense, and at the same time, there were rumblings of a shakeout beginning to occur in the electronic calculator marketplace. These factors contributed to a decline in profits at Electronic Arrays that began with a recession in the US, and steadily worsened over time due Electronic Arrays failing to diversity its IC business beyond low-margin devices such as Read-Only Memory ICs and electronic calculator chipsets. Along with that, the company was the falling behind in calculator chip development, with intense competition from Japanese chipmakers, as well as Mostek and Texas Instrumnets in the US. The financial issues mounted, and by 1978 the company was struggling with rather serious money problems, and were acquired by NEC Electronics(Japan) in 1978 in order for NEC to develop a chip manufacturing presence in the US. Along the rocky road to Electronic Arrays' failure and buyout, and about the time that the electronic calculator market suffered it's first big shakeout, the ICM subsidiary was liquidated. The remaining inventory of ICM calculators were later sold off through a surplus liquidator in San Jose, CA. ICM 816's were sold for as low as $135 each. It is not at all clear if ICM ever marketed any calculator other than the ICM-816. If anyone out there knows anything more about ICM and calculators it may have manufactured, the museum would greatly appreciate hearing from you. You can contact us by clicking the EMail box at the top of the exhibit page.
Inside the ICM 816
Another interesting and unusual machine in the museum has a similarity to the ICM 816, the Master H-1. The machines operate in a very similar manner, and in fact, share similar logic, both based on LSI chipsets made by Electronic Arrays. The The Master H-1 uses a more highly integrated four-chip chipset made by electronic arrays, that seems to have evolved within a fairly short period of time from the introduction of the S-100 chipset. Based on date codes found on chips, there seems to be about a one-year period of time from the introduction of the six-chip S-100 chipset, until a four-chip version with similar functionality was available.
Close-up View of ICM 816 Keyboard
The ICM 816 uses an 8-digit display to provide 16 digits of integer capacity. A special key (a double-ended arrow, designated as [<->] in this text) on the keyboard toggles the display between the most and least significant eight digits of results which exceed eight digits. For example, multiplying 12345679 by 18 would result in "22222222" showing up on the display. The clue that there is more of the result to be displayed is that there is no decimal point in the display. (If the result were actually 22222222, it would be displayed as "22222222.".) Since there is no decimal point shown, pressing the [<->] key changes the display to "2 ", showing the least significant digit of the result. Subsequent presses of the [<->] key toggles the display back and forth between the most significant and least significant digits of the result. If there aren't any additional digits to be displayed, pressing the [<->] key results in a blanked display.
Though the ICM 816 and the Master H-1 share a similar logic design, there are some subtle differences in their operation. The ICM-816 will only allow entry of up to eight digits of input. If more than eight digits of entry are attempted, the machine goes into input overflow state, requiring a press of the [CE] or [C] key to clear the invalid entry. The Master H-1 allows input of up to 16 digits (with the least significant eight digits not visible to the user while they are being entered). Another subtle difference between the machines is that the Master H-1 ignores attempts to set the decimal point position to 8 or 9, wheras the ICM 816 seems to get somewhat confused by doing this.
The ICM 816 uses fixed decimal point logic, with a setting from zero through seven digits behind the decimal point. Setting the decimal point location is not intuitive, as there is no obvious switch or dial provided to make the setting. At power-up, the calculator defaults to zero digits behind the decimal point, making the machine effectively an integer-only calculator. Based on experience with the Master H-1, pressing and holding the [CE] key, while at the same time, pressing a digit from 0 to 7 on the keyboard, properly sets the fixed decimal point location. Pressing 8 or 9 as the decimal point position selection is accepted, but results in the machine blanking the display, and immediately causing an input overflow as soon as any non-zero digit is entered. This method of setting the fixed decimal point position is very much like that used on the Marchant Cogito 412 and Cogito 414 calculators.
Closer view of ICM 816 LSI's
The US Patent (Number 3,800,129) filed by Electronic Arrays indicates that the chipset is a microcoded processor, utilizing ROM to provide the microcoded instructions that orchestrate the operation of other chips in the system to provide the brains for the calculator. The devices include an input chip (scan and encode keyboard), an output chip (multiplex and generate display signals), an arithmetic chip (perform logical and arithmetic operations), a register chip (provides three working storage registers and data routing), a control logic array chip (interprets microcode and generates signals to orchestrate the operation of the other chips), and a microcode ROM chip that contains the the microcode, as well as addressing logic for the ROM. The chipset IC's are numbered "110-5001" (Register Array), "140-5004" (Input Processing), "150-5005" (Output Array), "120-5013" (Control Logic), "160-5014" (Microcode ROM), and "150-5017" (Arithmetic Logic). The machine's construction is of high quality, with individually replaceable keyswitch modules (using magnetic reed switches for reliability), high-quality silkscreened fiberglass circuit boards, and overall very durable construction.
The Display Panel (note negative indicator lit at left end of display)
The ICM 816 uses eight genuine Burroughs B5853ST Nixie tubes for its display. Each tube contains the digits zero through nine, and a right-hand decimal point. The display is driven by an unknown IC (no part number or other identification on it at all) in a 14-pin DIP package, as well as discrete transistor drivers. The display provides leading zero suppression. The Nixies tubes sit behind a red filter that tints the orange Neon-glow display of the Nixies more toward the red end of the spectrum. Sign and overflow conditions are indicated by small discrete neon tubes located at the left (NEGative) and right (OVerFlow) ends of the bank of Nixie tubes.
The calculator is good about catching error conditions. Division by zero results in an immediate "OVF" condition, causing keyboard input to be ignored, requiring a press of the [C] key clear the machine and allow keyboard input. Any result that would exceed the 16-digit integer capacity of the machine causes the "OVF" indicator to light and the keyboard to be ignored (except the [C] key, of course). Entry overflow (typing in more digits than the machine can handle) results in an overflow condition that can be cleared with the [CE]" key. /FONT>
The ICM 816 is quite fast, with 99999999 divided by one requiring about a blink of an eye to complete. 99999999 squared (99999999 X 99999999) takes just a shade longer, but certainly less than 1/10th of a second.