Wang 720C Advanced Programming Calculator
After searching for quite some time for one of these wonderful pinnacles of Wang Laboratories' calculator technology, someone turned me on to not one, but two! Sadly, these machines had been left to languish in an outdoor storage area for years after they were taken out of service, and they were subject to the some of the ravages of the elements. The storage area was covered, but didn't have walls, and thus, the machines, while kept out of the rain, were subject to humidity, bugs, dust/dirt, cobwebs, and mice. The machines were rescued from this situation by a scrap dealer, who very fortunately realized that these were too cool to scrap, and, as luck would have it, contacted me. Even though the machines were in pretty rough condition, with two machines, it was possible to piece together and perform restoration of this machine to complete and fully operational condition.
Later, due to the kindness of Janet Harrison and other members of her family, another Wang 720C, along with a great deal of documentation and other materials, and a Wang 711 I/O Writer were donated to the museum. This equipment was owned by her father, Thomas Harrison, who passed away in 1996. Mr. Harrison used the Wang 720C in his mechanical engineering consultancy as a high-tech leg up on competitors that had to rely on simple desktop calculators. Sincere thanks to Janet and her family for making Mr. Harrison's prized equipment a lasting exhibit in the Old Calculator Web Museum.
The Wang 700-series calculators were the high-point of Wang Laboratories' electronic calculators. After releasing the 700-series, updated versions of the architecture were created; the business-oriented 600-series calculators, and the 'low-cost' scientific alternative to the expensive 700-series machines, the 500-series. As the 1970's progressed, the calculator business changed dramatically with the advent of LSI (Large Scale Integration) integrated circuits. LSI chipsets condensed the entire functionality of a scientific calculator down to a few IC packages. Handheld calculators were becoming prolific, and margins in the calculator business were dwindling. Dr. Wang realized that his company could not make itself a force in the semiconductor business soon enough to make a difference, and that, as LSI technology took over, his calculator business would be at the mercy of the chipmakers. So, he decided to gradually phase Wang Labs out of the calculator business, concentrating the resources of the company on word processing and small business computer systems. Wang continued to sell the 700-series calculators and their predecessors into the late 1970's, but by that point, handheld calculator technology could replace much if not all of the functionality of such desktop behemoths.
The Elektronika D3-28
Image Courtesy Konstantin Zeldovich
Before going into the history of the Wang 700-series calculators, it's very worthwhile to relate an interesting story about a Soviet-made 'clone' of the Wang 700. It is said that imitation is the most sincere form of flattery. Well, the Wang 700 proves this to be true. Apparently, somewhere in the Soviet Union, the Wang 700 caught the attention of some authority involved in Soviet technology, and the machine was thought highly enough that they decided to build a machine of their own that matched the 700. This machine was called the Electronika D3-28. The Soviet-built calculator is, for all intents and purposes, a replica of the Wang 700, but packaged in a much more sleek, refined, compact, and service-friendly cabinet, with a cabinet design reminiscent of the Hewlett Packard 9810.
Profile View of the Elektronika D3-28
Image Courtesy Konstantin Zeldovich
The D3-28 was introduced in the late 1970's (seven or eight years after Wang's introduction of the 700-series calculators), and was produced well into the early 1980's. The machine pictured here appears to have been manufactured in 1981, based on a date listed on the model/serial number tag, as well as from date codes on various parts within the machine. The D3-28 utilizes very similar technology to the Wang 700, including a virtual clone of Wang's unique ferrite transformer ROM for microcode storage, and small to medium-scale Integrated Circuit design. The IC's used in the machine are of the Soviet K155 family, which are very similar to US-made 7400-series TTL devices. In some cases, the K155 devices are even pin-for-pin compatible with 7400-series devices. The only real departure in the design of the D3-28 is the use of integrated circuit RAM chips (K565RU1A memory devices) as opposed to the magnetic core memory used in the Wang 700-series calculators. By the time the Soviet version was introduced, integrated circuit memory was more cost-effective than core-memory technology. The microcode ROM of the D3-28 has the same number of bits as the Wang 700, indicating that perhaps the microcode is a clone of the microcode for the 700. If the microcode is a copy of the Wang 700 microcode, this would be a certain indication that architecturally, the D3-28 is virtually identical to Wang's machine. There are some signs that the microcode, while very similar, is not an exact copy of the Wang microcode. An example of a subtle difference is that the D3-28 provides five digits for the program counter display when in "LEARN" mode, versus four digits on the Wang 720C. It is possible that the Soviet machine, by virtue of its solid-state memory, may have subtle microcode changes to allow it to address more memory than the 700-series machines.
The Ferrite Transformer ROM of the Elektronika D3-28 (left). Compare to the ROM in the Wang 720C (right)
D3-28 ROM Image Courtesy Konstantin Zeldovich
It seems extremely likely that the Soviets got ahold of a Wang 700 and reverse-engineered it, and in addition, perhaps they may have obtained of a copy of the Wang 700-series service manual (which contains schematics), and from this information, went about re-implementing the logic to match their integrated circuit technology. The D3-28 uses two seven-segment gas-discharge display panels (similar to Burroughs Panaplex devices) rather than the dual-row Nixie tube display of the Wang 700-series machines. The Soviet machine, like the Wang 700-series calculators, provides a cassette tape drive for storing and retrieval of programs and data. The keyboard design is quite different than that of the Wang 700 calculators, with the keyboard using more conventional keyswitches rather than the microswitch design that Wang used. Also, the various mode control switches are integrated into the keyboard (with LED indicators to show which mode the machine is in). Like the Wang 700-series, the area above the special function/addressing keys is provided for strips which can be used to provide user-defined labels for the keys.
Profile View of the Wang 720C
The story behind the Wang 700-series calculators is a story that repeats itself time and time again in the realm of high technology business -- the story of vaporware. What became the 700-series calcuators was really not intended to be a calculator. On top of the high-end calculator industry, Dr. Wang felt that his company could compete in the commercial computer market; a market that was dominated mainly by IBM, along with Wang's "neighbor" located in Maynard, Massachusetts, DEC (Digital Equipment Corp.), and a smattering of other players such as General Electric, RCA, Burroughs, and Control Data. Wang wanted his company to build a computer that could compete with IBM's wildly successful IBM System/360. Dr. Wang wanted a computer that Wang enter into the marketplace that would be less expensive than IBM's computers, yet provide the same type of functionality. Wang Laboratories had a major cash cow with the extremely successful 300-series electronic calculators (an example being the Wang 360SE), and had plenty of money to fund the development of a computer to compete with IBM. However, the 300-series calculators were a design that was created in the mid-1960's, and the technology of the machines was rapidly becoming dated. Electronics technology had changed quite dramatically during the few short years between the design of the 300-series calculators and early 1968 when the design of Wang's computer system began. Integrated circuit technology had started to become a reality, which made it possible to jam more complex circuitry into a significantly smaller package. While Wang Labs was resting on the laurels of the success of the 300-series, spending lots of money developing a computer to compete with IBM, other established electronics companies as well as various upstart entrepreneurs were realizing that IC technology could be used to build calculators that could be used to steal lucrative market share from Wang.
The rather intimidating control panel of the 720C
It was March of 1968, and the design of Wang's new computer system was moving along at a good pace. Dr. Wang and some of his senior folks went to New York City to the IEEE Electro Conference being held there. Little did Dr. Wang know that a machine was being demonstrated in a private suite at this event that would put Dr. Wang and his company in a panic. Bill Hewlett, one of the founders of Hewlett Packard, and a man whom Dr. Wang was familiar with, invited Dr. Wang to come see "something" in one of their private show suites. In the room was one of a limited number of pre-production prototypes of the HP 9100A electronic calculator. Dr. Wang, Bill Hewlett, Barney Oliver (9100 Project Leader), and Tom Obsbore (Chief 9100 Architect Contractor) were there when Dr. Wang got his chance to see the 9100A. After a few minutes of seeing the amazing speed, power, and accuracy of the 9100, Dr. Wang thanked Bill Hewlett for showing him HP's latest brainchild, and let the room saying that he must "get to work". The realization that Dr. Wang felt at the moment was that the 9100A made Wang's 300-series calculators look like dinosaurs in comparison. Even the programmable Model 370 and 380 Wang calculators simply did not have the features, speed, and functionality that the 9100A (and a later, the even more capable 9100B) had going for it. The 9100 was indeed a market breaker, and Dr. Wang knew something had to be done to counter it, and fast, or Wang's dominance in the high-end calculator market would soon end.
Even though the HP 9100's were based on transistorized technology like the 300-series Wang calculators, the HP machine leveraged an amazingly innovative microcoded design (a concept previously applied to the design of computers) to wring the ultimate computing power out of the size limits of a desktop package. Shortly after HP formally announced availability of the the 9100A calculator in September of 1968, sales of Wang's 300-series machines started dropping dramatically. But, thanks to Bill Hewlett's "wake up call", moves were already afoot within Wang Laboratories to set the tables straight. Dr. Wang went to the folks designing the new computer and told them to turn the design into a high-end programmable calculator. Dr. Wang's reasoning was that any computer can be made to act like a calculator by just changing the programming, which was true, and comparatively trivial to implement (as opposed to redesigning hard-coded logic). In fact, the computer being designed was to use a microcoded architecture similar to that of the IBM 360 computer. Micro-coding uses a ROM (Read Only Memory) to store very low-level instructions that guide the transfers of data and operations performed on data within a generalized set of registers and logic units. Since the computer being designed was based on this microcoded architecture, all that was needed was some additional hardware (Nixie tube display system, specialized keyboard, and other calculator-specific hardware changes), scale down the microcode to control a smaller calculator-specific register and processing architecture, and write the new microcode to make the system operate as a calculator rather than a computer. Dr. Wang ran a very monolithic company; essentially, everyone reported to Dr. Wang, even though there were other reporting structures in place, everyone had a dotted-line reporting structure to Wang. This meant that whenever Dr. Wang had a big project in mind, anyone's priorities could very quickly change. It's clear that Dr. Wang's preview of the 9100A in New York quickly changed his mind about trying to do battle with IBM, and focus more on saving his business - and indeed, Wang's visions of conquering IBM faded quickly from the forefront of his mind (though they never really did go away).
The principle hardware designer for the "IBM-beater" computer project at Wang was a man named Dave Moros. Moros had come to Wang Labs as the result of Wang's acquisition of a company called Philip Hankins, Inc., or PHI for short. PHI was a programming and computer service bureau with it's own IBM System 360/65. Moros was a skilled programmer, but proved to be very adaptable, and quickly understood the hardware side of what made computers tick. Moros was retasked to convert the logic design for the computer into a high-end calculator. Mr. Shu-Kuang Ho and Mr. Don Dunning were tasked with taking the logic designed by Moros and turning it into real hardware. With these key people in place, the hardware side of the new calculator project was staffed.
One critical resource needed to get this calculator project going was missing -- the microcode to make the calculator run. Dr. Wang pitted some of his most brilliant engineers against each other in a high-pressure, tight-deadline contest to very quickly generate the most efficient microcoded math algorithms for the new computer-turned-calculator.
As opposed to most computers of the time, a high-end calculator must be able to operate on floating-point decimal numbers, and, along with the four basic math functions, must also be able to perform more complex math operations such as logarithms and square root. On top of this, the calculator must fit on a desktop. These requirements add a level of complexity to calculator microcode that can be typically be ignored when designing a computer. As a result, extremely efficient microcode was needed to be able to implement all of the functions the calculator required, and still allow the code to fit within the limited capacity of the transformer-based ROM technology that Wang had developed as the microcode store for the computer. Along with the microcode size constraints, the microcode algorithms themselves had to be very efficient, allowing the new calculator to be generate answers as fast as or faster than the competition, yet still retain the high level of accuracy that Wang calculators were known for.
Harold Koplow, August, 1968
One of the participants in this contest was a young engineer named Harold Koplow. Koplow had been hired on at Wang Labs as a programmer writing applications for Wang's Model 370 programmer for the 300-series calculators. Koplow seemed to have an intuitive understanding of how to wring the most capability out of the limited capabilities of Wang 370...which seemed just the skillset needed to design the microcode for the new calculator. Koplow ended up winning the contest, designing microcode that took the fewest microcode instructions to implement a given function.
Koplow's brilliance and perserverance in the development of the microcode for the Wang 700-series calculators helped to vault him into the role as Wang's senior calculator engineer after the 700-series project was finished. After the 700-series project, Koplow was a primary figured in the development of Wang 500, 600 and 100-series calculators. Later, Koplow was instrumental in the creation of Wang's word-processing systems, which became Wang's big "next generation" cash-cow business after the company left the calculator business toward the end of calculator market shakeout of the mid-1970's.
Cover of February 24, 1969 Edition of "Product Engineering" Magazine
In December, 1968, Wang Labs announced the new 700-series calculator to compete directly with HP's 9100 calculators. As many people who know high-tech industry understand, the announcement of a product doesn't necessarily mean it exists. Even back in late 1968, the notion of "vaporware" existed, and Wang's introduction of the 700-series was just that. Even though the product had been announced, shipments weren't promised to begin until mid-1969, which, as it turned out, was a milestone that Wang missed by a few months. In February of 1969, McGraw-Hill Publishing's Product Engineering magazine published a cover-story on the new (and not yet available) Wang 700, featuring a proud, cigar-toting Dr. Wang surrounded by a number of 300-series keyboard/display units and his new baby, a prototype Wang 700. Also in the February 1970, Wang Laboratories' monthly publication, the Wang Laboratories "Programmer" published an announcement article for the Wang 700, not so subtly bragging that the machine was the fastest and most capable machine on the market.
Early Prototype Wang 700-Series Calculator
Image Courtesy Bill Needler
By April of 1969, Wang was distributing slick marketing materials touting the new machine, with indirect but rather obvious comparisons between the new 700, and Hewlett Packard's 9100. This early marketing literature was clearly developed from engineering specifications and prototypes, and didn't accurately reflect what turned out to be the actual product. An example of this is in a section of the literature explaining the performance of the calculator. It is stated, "The 700 is the fastest by far.", with performance figures quoting an addition or subtration in 300 microseconds, multiplication in 3 milliseconds, logarithm in 15 milliseconds, and ex in 35 milliseconds. The actual performance the production calculator delivered were substantially slower (though still quite fast), with addition/subtration in 1.7 milliseconds [almost 6 times slower than quoted], multiplication in 12 milliseconds [4 times slower], logarithm in 47.2 milliseconds [about 3 times slower], and ex in 61.8 milliseconds [just under half as fast]. The changes in the speed of the machine compared to early marketing literature are likely the result of aggressive marketing-driven specifications which had to be backed off when the reality of the electronic implementation became clear. Even so, the 700-series calculators are placed among the fastest high-end electronic desktop calculators of the era. Even today the Wang 700-series calculators are superior in speed to some of today's advanced calculators.
The prototype machines shown in the early marketing materials and the press coverage looked similar to the final production unit, but a number of differences stand out. Most notably, the cassette deck in the prototype was positioned differently, with horizontal cassette loading rather than the vertical loading that came to be in the production unit. It almost appears that a consumer-grade cassette recorder was adapted to fit in the unit. This was replaced by a totally different mechanism in the production units. There were also subtle differences in the keyboard layout, with the most obvious being the absence of the pushbutton switches controlling the mode of the calculator. Also, the prototype did not have the large fan on the back of the cabinet...a later add-on that was necessary due to the electronics of the machine generating more heat than could be cooled by simple convection.
Later Prototype Calculator with
Vertical Loading Cassette Drive
Image Courtesy Bill Needler
A later advertisement for the 700-series shows the changeover to the vertical-loading cassette drive, but with a bezel-color (black) door rather than the final beige cassette door used on the production machines.
The delays that Wang experienced in getting the 700 calculator into production were a benefit to competitors, especially Hewlett Packard. The time between Wang's announcement of the new machine and the time that the machine was actually shippable was well used by Hewlett Packard and others lure more sales away from Wang's bread-and-butter calculator business. The engineers working on Wang's computer project were put on a drop-dead schedule to meet the promised mid-'69 first customer ship deadline. As it turned out, the deadline was missed, but not by much.
Built-in Cassette Tape Drive for Program Storage
Computer systems of the time were not desktop devices, usually taking up space in an equipment rack. The design for the computer that Wang's engineers were working on wasn't really intended to fit on a desktop. The computer was designed to be integrated into a desk, with the area where the file drawers would be on a regular desk packed with the electronics that make up the computer. This design needed to be shrunk down quite dramatically to make it into a desktop package. There simply wasn't enough time. When a big electronics industry trade show where the 700 was to make its debut rolled around, eager customers were expecting to see Wang's wonderful new machine that had been marketed strongly and praised by the press. Some devout Wang customers had already placed advance orders for the machine. The problem: It wasn't ready! A prototype of the electronics of the calculator existed and worked well, however, the electronics were still too large to fit inside the package that had been designed for the desktop calculator. So, the engineers put the keyboard, display, and cassette deck into one of the cabinet mock-ups for the 700-series machine, bolted that cabinet to a table-top, and fed a big bundle of cables out of the bottom of the cabinet, through a hole in the table top, to a hidden area under the table that contained all of the electronics of the calculator. What the attendees of the trade shows saw was a very powerful and advanced calculator that, while rather large, still fit on a desktop. What they didn't know was that the rather chunky looking cabinet they saw contained mostly air. The deception was a success, at least for a while. Orders for the 700-series machines started pouring in. While orders are great, sales are what keeps a company going, and still, Wang had nothing to sell. Wang's salespeople were advised to placate irritated customers whose orders for 700-series calculators were not being filled by providing them with loaner high-end 300-series calculator systems until their orders for the 700-series machines could be fulfilled. Finally, slightly behind schedule in the 3rd quarter of 1969, the first production Wang Model 700 calculators were rolling off the assembly line.
Photo at a Trade Show in September, 1970, with Wang 700-Series Calculator(left) and Model 701 Output Writer(right).
Note no fan on the back of the 700-Series Cabinet
Photo at a Trade Show in September, 1970, with Wang 700-Series Calculator(left) and Model 701 Output Writer(right).
One of the biggest problems Wang had with the early 700-series calculators was cooling. The electronics were really designed to be in a computer cabinet rather than a comparatively small desktop calculator. As such, the early 700-Series calculators had some pretty serious problems with overheating.
Early Wang 700A with Convection Cooling (No Fan)
Image Courtesy Bill Needler
Early Wang 700A with Convection Cooling (No Fan)
Originally, the machines were designed for convection cooling, with vents in the base of the chassis and top of the cabinet. This didn't work well at all. In fact, Wang salespeople had to adopt an interesting, if bit misleading strategy to prevent potential customers from realizing that Wang had a problem with the cooling of the calculators. Salespeople demonstrating the machine at trade shows or customer sites learned quickly that the early 700-series demo calculators would start to exhibit strange symptoms after operating for around 15 minutes because they would overheat. So, they would demonstrate the amazing capabilities of the calculator for about ten minutes, then they would turn the machine off and pull off the top cabinet (fortunately, it was very easy to remove), showing off the high-tech innards of the machine, conviently allowing the machine to cool off while they were explaining the wonders of Wang's use of integrated circuit technology.
Later 700-Series Machines Used Forced Air Cooling
Later 700-Series Machines Used Forced Air Cooling
After a few minutes of showing off the insides of the machine, the machine cooled enough for the electronics to be happy again, the cover went back on, and the salesperson started the spiel all over again. It soon became clear that real customers wouldn't relish having to operate their machines without the cover, so a production change was implemented, involving the addition of a good-sized fan mounted on the back of the cabinet. The fan provided forced-air cooling for the electronics, eliminating the cooling problems, but somewhat corrupting the lines of the machine, not to mention adding fan noise to what before was a silent instrument.
Dr. Wang's redirection of his computer effort was a success, at least for a little while. The 700-series calculators did quite well in the marketplace, proving to be worthy competition for HP's 9100's. However, during the time that Wang was working on hammering out the 700-series, HP was busy designing a new line of calculators that would truly blur the line of definition between a computer and calculator...the 9800-series, a triple-whammy assault of powerful calculators that would spell the beginning of the end for Wang in the calculator business. The HP 9810A, introduced just a little more than a year after the Wang 700-series calculators began shipping, used large scale IC technology, and was more than a match for the 700-series machines. To complete the three strikes against Wang, shortly after the 9810A was introduced, HP announced the 9820A Algebraic calculator and 9830A, a computer-like calculator programmed in BASIC.
Under the cover of the Wang 720C.
The Wang 700-series machines underwent a number of revisions as the product line matured. The 720C featured here is the highest-end machine in the 700-series.
There are two main flavors of the 700-series calculators, the 700, and the 720. The Model 720 machines received double (16K bits) the amount of core memory capacity of Model 700 calculators. The original machine in the series was the 700A, with the 720A providing twice the memory capacity. The -A version machines were quickly superceded by the -B version machines. The -B version of the 700/720 involved some design improvements, with the major change being the addition of new microcode to the ROM to allow interfacing of the Model 701 Output Writer or 702 Plotting Output Writer devices. The -C versions were a later update, including some minor circuit board updates, but mainly incorporating additional microcode modifications to add some new operations to the calculator's instruction set. The -C microcode changes added math (mostly decimal point positioning) instructions, along with instructions to handle transfers to/from main memory from an external memory expansion interface (Model 708-1 External Memory Controller, and Model 708-2 External Memory Modules). Wang Labs designed the 700-series machines to be retrofittable, meaning that it was possible to upgrade a 700B to a 720C in the field by simply changing circuit boards, core memory, and ROM. The 720C exhibited here appears to have been a benefactor of such an upgrade, with a new model/serial number tag affixed over the top of the original Model 720B model/serial tag.
The Dual-Register Nixie Tube Display of the 720C
One of the most striking features of the 700-series calculators is the display. All of the machines in the series have a dual-register Nixie tube display...more Nixie tubes than any other Nixie calculator that I'm aware of. The display is arranged in two rows of 16 tubes each. 12 digits are used for normal numeric display, or for the mantissa on numbers displayed in scientific notation. A special sign Nixie (which actually contains a +, -, 8, and left and right-hand decimal points) is located at the left end of each row for indicating the sign of the number in the display (the 8 and left-hand decimal point in this tube are not used). Separated slightly and to the right of the mantissa display is the exponent display, consisting of another 'sign' tube, and two numeric tubes. The numeric Nixies contain the digits zero through nine, and both left-hand and right-hand decimal points, but only the right-hand decimal points are used. The tubes themselves are labeled as Wang parts, with Wang part numbers, but it's most likely that they were manufactured to Wang specifications by Burroughs.
One of the Nixie Driver Circuit Boards
The Nixie tubes plug into sockets mounted on circuit boards that connect to the Nixie driver circuit boards by a cable terminated with edge-connector fingers. The two arrays of Nixie tubes are mounted in an aluminum extrusion that provides a framework for holding everything in place. A foam-rubber strip across the tops of the tubes provide shock isolation and positioning for the tubes. Two identical driver circuit boards contain the circuitry to drive the Nixie tubes, including a TTL decoder chip, and discrete transistor driver circuitry. The Nixie display is multiplexed, meaning that each Nixie displays its digit in a timeshared fashion, with the timesharing occurring at a fast enough rate that the human eye perceives all of the Nixies illuminated at once.
The Nixie Tube Display Module
The bottom display register is called the "X" register. This register is where all numeric entry occurs, and results of scientific functions (for example, square root) are placed here also. The top register is the "Y" register. This register can be used in a number of ways, but has three major uses. First, it can get a copy of the number currently in the "X" register by the user pressing the [^] (up arrow) key. The Y register also acts as an accumulator, automatically accumulating the results of addition or subtraction functions. Lastly, the Y register displays the results of multiplication or division operations. The displays have fully floating decimal, and automatically switch to scientific notation when the number is too large to be displayed conventionally. Numbers are displayed left justified in the display, and trailing zeros are not inhibited. Numbers displayed in scientific notation are displayed in a somewhat unusual format, being left-justified, with the decimal point always before the first digit. For example what would normally be written as 5.0x10-15 would display as "+.500000000000 -14". You can see examples of the display in scientific notation in the photo above.
The backside of the Keyboard Key Assembly
Another distinguishing feature of the Wang 700-series machines is the vast keyboard on the machine. There are a total of 83 keys on the machine, and at first glimpse the sheer volume of keys can be intimidating. The keys are grouped by function, with further visual cues of color-coded keycaps to help the user quickly find the function they are looking for. Clear keycaps are generally related to numeric entry and control functions, blue keycaps indicate higher-level math functions and other specialized functions, rose-colored keycaps are related to data storage and retrieval, and smoke-gray keycaps designate programming and program control functions. The keys themselves are an unusual design. The design of the keys harkens back to the original Wang LOCI-2 electronic calculator, Wang's first entry into the marketplace. The only real difference in the design of the keyboard between the LOCI-2 and the 700-series machine is that the keycaps on the 700-series machine are smaller, due to the sheer volume of keys on the 700. The keys consist of a flat plastic square sitting on the top end of a stalk. Over the top of the square, a paper legend containing the nomenclature for the key is placed, then covered with a snap-on plastic cover that retains the nomenclature on the keycap. The snap-on cover is made of clear (or tinted) plastic to allow the keycap nomenclature to show through.
The Keyboard Circuit Board, with myriad Microswitches
The stalk of the key is square, and fits through a square hole in the keyboard panel. On the back-side of the keyboard panel, a round plastic disc (looks like a plastic washer) is pressed onto the stalk, such that the disc is perpendicular to the stalk. This disc rides on the plunger of a circuit board-mounted microswitch, such that when the switch is pressed, the plunger of the microswitch is depressed, activating the switch. The spring pressure of the microswitch pushes the keycap back up when the key is released. This makes for a keyboard that has very little key travel, but works surprisingly well, is mechanically very simple, and is extremely reliable. The keys on the 720C in the museum work as well today as they did when the machine was brand new, with no bounce or false entries.
The Bare Wang 720 Chassis
Moving inside the machine, it's clear that Wang designed the machine to be modular. Most of the guts of the machine, except for the microcode ROM array, are situated on a sturdy thick-gauge aluminum chassis. The chassis contains the edge-connectors that the circuit boards plug into, the power supply (which is also modular), and mounting points for the Nixie tube display subsystem and the cassette tape drive. The back panel of the chassis is exposed through a cutout in the back portion of the lower half of the case of the machine, and has connectors for I/O expansion and a printer, along with a fuse holder, power switch, and strain-relief for the power cord.
The Hand-wired Backplane of the Wang 720C
It seemed that Wang Labs wasn't afraid of building machines that were labor intensive to make. Many of Wang's machines used hand-wired backplanes to interconnect the circuit board, and the 720C was no exception. The backplane is a maze of multicolored wire strung point-to-point between the long pins of the edge connectors. Most of the backplane wires connect to the edge connectors with clips. The edge connectors have long square-pin tails which the clips connect to. The clip forms both a mechanical and electrical connection between the wire and the edge connector. The edge connectors are framed by a large circuit board with cutouts in it so the edge connector pins can stick through. This circuit board serves as a power distribution plane for routing the power supply voltages needed for most of the logic in the machine.
The Circuit Boards of the 720C
The majority of the logic of the 720C is contained on compliment of twelve main circuit boards with a combination of small-scale DTL (Diode-Transistor Logic) and TTL (Transistor-Transistor Logic) integrated circuits. A total of 315 IC's make up the logic of the machine, along with a large number of transistors, diodes, and other discrete components. The TTL parts are all standard 7400-series devices, however, many devices have Wang-proprietary part numbers on them of the form 376-xxxx. It appears that Wang used these internal part numbers on standard devices in order to keep the service and parts sales for these machines to their own service organization. Along with that, Wang probably wanted to make it more difficult for competitors to reverse-engineer the machine. It appears in one case that one of Wang's IC vendors (Texas Instruments) made a mistake on labeling some of the parts, including both the Wang internal part number, along with the standard 7400-series part number. Wang had to paint over the 7400-series part number to 'keep their secret'. In any case, the secret wasn't kept all that well, because when comparing identical boards between the two machines, there are cases where there are Wang-numbered parts that correspond to standard parts in the same locations on comparable circuit boards. My guess is that Wang couldn't get enough volume of the custom-numbered parts, and had to resort to using 'off the shelf' parts in order to meet manufacturing demand.
Closer view of one of the Wang 720C Circuit Boards (Note rectangular painted-over areas on some of the parts)
One problem with IC technology of the late 1960's is that it was great at integrating logic, such as gates and flip-flops, but the technology for providing memory, such as RAM (Random Access Memory) and ROM (Read-Only Memory) simply didn't exist. The computer that Wang originally set out to design was to be a microcoded architecture. Microcoded architectures use generalized logic elements that are controlled by sequencing electronics that read 'microinstructions' out of ROM and direct the operation of the logic. A microcoded architecture allows much more generalized logic to perform a myriad of tasks. The microcode for a processor architecture is essentially a set of 'instructions' that direct the operations of the logic of the machine.
The Ferrite Transformer Microcode Store of the Wang 720C (Component Side w/Signal Steering Diode Array)
Microcoded architectures are significantly more flexible than hard-wired logic, in that changes to the behavior of the logic can be made by simply changing the microcode. For example, later electronic calculators, such as those made by Compucorp, used the same basic logic architecture, with microcode changes giving each different model a different set of functionality. The problem with microcoded designs was that the microcode needed to be stored in a read-only form somewhere. At the time, integrated circuit ROMs simply didn't exist. So, Wang had to build their own ROM. Earlier Wang calculators used diode-based ROMs for built in 'programs' to perform trigonometric functions. However, a diode-ROM that could hold the amount of data required for the microcode for such a complex machine would be prohibitive. So, Dr. Wang ended up using technology that he had a hand in the development of -- a ferrite rod-based ROM.
The "Wire" side of the Wang 720C Microcode ROM
The microcode ROM is a good example of Wang Labs' fearless attitude with regard to labor-intensive designs. While the component side of the ROM board is interesting, with literally thousands of tiny diodes (used for routing signals through the wiring of the ROM) soldered into place. While tedious, automated equipment could be used to place and solder these diodes. The other side of the circuit board is where the labor intensive part comes into play. Literally thousands of hair-like strands of enamled magnet wire are strung acorss the board, stretching from the diode selection logic up through wire guides, and into sqaure-shaped plastic fixtures at the top of the board. In early 700-series calculator prototypes, each of these tiny wires had to be hand-threaded, a task that used over 5000 feet of enameled wire, and took six and a half weeks to complete. Obviously, manually wiring the ROM was out of the question. Dr. G. Y. Chu, one of Dr. Wang's closest friends, and Wang's senior staff engineer, devised a semi-automatic "weaving machine" that assisted a patient operator in the process of wiring the ROM. While this machine significantly reduced the time required to manufacture a ROM, it still took too long.
The Production Wang 700 ROM "Weaving" Machine
Photo from February, 1970 Wang Laboratories Programmer Periodical
In the end, the weaving machine was augmented with additional electronics, and, interestingly enough, had its operation controlled by a ROM very similar the one the weaving machine was to manufacture. This augmentation allowed a ROM to be manufactured in an automated fashion in a period of just under one day. Dr. Chu ended up getting a US patent (3639965) on this amazing machine that was granted in February of 1972. A bunch of these automated weaving machines were quickly constructed, allowing volume production of the ROMs for the 700-series calculators to begin. The difficulties of creating the ROM for the 700-series machines were the primary reason behind the delays in the delivery of Wang 700 calculators to customers.
A close-up of the Hand-Wired connections on the 720C Microcode ROM
The active part of the ROM consists of eleven square plastic structures, each of which contains four U-shaped ferrite structures. Each of these ferrite structures effectively serve as small transformers, with primary and secondary windings. A winding of wire is wrapped around one leg of the U shape, connecting to a transistorized sense amplifier. This winding is the secondary winding of the transformer. The other leg of the "U" has the primary windings.
A close-up of the Ferrite transformer elements and windings
If one of the wires wrapped around the primary side of the "U" carries a pulse of current, that pulse will be picked up in the secondary winding, be amplified by a sense amplifier, and sent off to the microcode logic. A total of 43 of these ferrite structures are used to make up the 43-bit microcode control word that drives the operation of the Wang 720C. An array of TTL decoder integrated circuits, along with the gigantic array of diodes serve to decode and select the microcode address into one-of-2048 individual word lines. When a given word line is decoded, a pulse of current flows into one of the 2048 tiny wires. The selected word line wire threads its way from the selection logic, through some plastic wire guides, then up through the array of ferrite rods, wrapping around a rod when a '1' is to be encoded for that bit, and bypassing the rod when a '0' is to be encoded. A similar type of technology, using toriodial-shaped ferrite elements, was used for microcode ROM in the Hewlett Packard 9100 series calculators, the machines that pushed Wang to build the 700-series calculators.
The microcode ROM is organized as 2048 words of 43 bits each, for a total of 88064 bits. Given that the circuit board is approximately 16" x 12", that comes to a bit density of just over 458 bits per square inch. Today, we can cram 128 million bits of ROM onto a chip of silicon a little smaller than the fingernail on your pinky finger.
720C ROM Board in place in the base of the cabinet
The wire side of the ROM board is covered by a moulded plastic sheet that is taped in place to protect the delicate wiring from damage. It is fortunate that Wang decided to provide this shield from the elements, as it is sure that the wiring would have been damaged by the critters that inhabited these machines while they were in storage. The ROM microcode board occupies a good portion of the base of the machine, mounted in the base with rubber isolating pads to prevent shocks from handling from jarring the wiring. The ROM board connects into the main chassis of the calculator by two edge-connectors.
Wang 720C Core Memory Board
Along with having to deal with the limitations of technology in the microcode ROM, Wang also needed to build machine with a significant amount of random-access memory to store user-written programs and memory registers for data storage. At the time, integrated circuit memory chips were just beginning to be experimented with in the laboratories of IC manufacturers. So, Wang relied on the tried and true memory technology that Dr. Wang helped to invent...core memory. All of Wang's earlier calculators utilized core memory to store memory and working registers, and the 700-series machine was no exception. The 700-series machines were Wang's last to use core memory. The later follow-on 500 and 600-series Wang calculators used the same architecture as the 700-series machines, but replaced the core memory with MOS (Metal-Oxide Semiconductor) memory chips. The core memory plane in the 720C consists of 16384 cores, arranged in 8 planes. The cores are strung on tiny gold wires. Core memory uses magnetic fields to store a single bit in each core. The nice property of magnetic core as opposed to semiconductor memory is that core memory retains its state even when power is removed, meaning that program steps and data stored into the 720C's memory years ago was immediately retrievable once the machine was brought back to life. The core memory board itself contains only the core arrays and diode switching circuitry. A clear plastic shield is secured to the board to provide protection for the delicate core planes. Another board provides the various driver, timing, and interface circuitry to allow the memory to interface with the rest of the calculator logic.
Display with 720C in "Learn" Mode
The 720C is a very capable calculator. In addition to the usual four math functions, the machine provides single-key solutions to square root, x2, 10x, ex, base 10 and e Logarithms, reciprocal, absolute value, and integer functions. A [Pi] key immediately recalls the constant Pi into the X register. The machine also provides extremely powerful memory register functions, allowing both direct and indirect access to memory registers. Individual function keys provide access to directly or indirectly addressed register math (add, subtract, multiply, and divide) between the X register and a given memory register. Memory registers can also be stored into and recalled directly and indirectly (with the memory register defined by the number in the Y register during indirect operations), as well as swapped between the X register and memory register. An unusual feature of the 700-series machines is the [RECALL RESIDUE] key, which enables easy multiple precision calculations to take place. In cases where results of addition, subtraction or multiplication exceed the 12 significant digit display capacity of the machine, this key recalls the extra digits of the answer into the X register for use in further calculations. In division problems, the remainder after the division is performed is recalled into the X register by the [RECALL RESIDUE] key.
The 700-series machines are also very
powerful programmable instruments. Programming is performed using
"learn mode", where keystrokes are translated into codes that are stored
into program memory one at a time. Each keystroke consumes one 'step' of
program memory, and is represented in the machine as two two-digit
decimal numbers between 00 and 15. For example, the  key is represented
with the code "07 02". For a listing of the various operation codes
on the 700-series calculators CLICK HERE.
When the calculator is in learn mode, the Y register display is blanked,
and the X register display shows the current program memory location
followed by the program code at that location. Programs
can do extensive test and branch instructions, allowing looping and decision
making operations to be performed based on the results of calculations or
user input. Branching operations are performed by marking a branch
destination with a label (by pressing the [MARK] key, followed by any
other key to serve as the label), with branches to that location caused
by invoking the [SEARCH] key followed by the label to be branched to.
Conditional test operations perform the specified test
(such as Y
An Early Wang Cassette Tape for Wang 700-Series Calculators
Image Courtesy Bill Needler
A Later Wang Cassette Tape
Item Courtesy of the Thomas J. Harrison Family
The 700-series calculators have a built-in cassette tape drive that allows programs to be easily stored and recalled from cassette tapes. The drive is semi-automatic, requiring that the user properly position the tape with manual "FORWARD" and "REWIND" controls, and making the tape ready for access by pressing the "TAPE READY" control. The "RELEASE" control ejects the tape.
Wang 700-Series Program Library Binder
Item Courtesy of the Thomas J. Harrison Family
Wang 700-Series Program Library Binder
The value of any programmable calculator comes in its ability to be adapted to a wide variety of tasks, depending on its programming. Wang realized early in their days in the calculator business that it was a worthwhile effort to provide a library of programs that their customers could leverage so that they didn't have to end up writing programs themselves. Wang Labs provided an extensive library of programs which were made available to any purchaser of a 700-Series calculator. The library (as it existed in 1970, when the example of the library in the museum was copyrighted) consisted of programs in eight different categories, including Mathematics, Civil Engineering, Statistics, Mechanical Engineering, General Science, Business and Finance, Electrical Engineering, and Utility. Each program in the library was documented with a program description, abstract, operating procedure, examples, a complete program listing, and in some cases, supporting documentation such as flow charts.
Given the powerful programming capabilities of the machines, it is only natural that the 700-series calculators also be able to interface with a wide variety of external equipment. Wang designed the 700-series calculators with a generalized input/output system that allowed a wide range of peripheral devices to be connected to the calculator. These peripherals included printers and teletype units, external magnetic tape drives, paper tape readers, and digital I/O ports that allowed the 700-series calculator to interface with just about any kind of digital device. Wang published a detailed document outlining the I/O interface of the machine, allowing customers to design their own I/O interfaces, making the 700-series calculators extremely versatile data acquisition and control devices. An example of this flexibility is that NASA used a 700-series calculator during the Apollo spaceflights as a range safety system to track lightning strikes and warn flight controllers of unsafe atmospheric electrical conditions during pre-launch activities.
The 700-series calculators are rather fast. All operations return results that appear instantaneous to a human operator. A printed specification sheet for the 700C lists approximate timing for various calculations, with addition/subtraction completing in 300 to 500 microseconds, multiplication and division in 2 to 5 milliseconds, and square root (the slowest operation) completing in 25 milliseconds. While calculations are occurring, the displays flicker ever so slightly. The circuit board that appears to control the timing of the machine has an 8.0 MHz crystal on it, implying that the master clock frequency of the machine is 8 Megahertz. This master clock is likely divided down to some even fraction of 8 MHz, probably something in the range of 500 KHz to 2 MHz, given the type of logic used in the machine. Whatever the basic operational speed of the machine, it is fast enough that even complex programs execute at a fair clip. A simple program loop that continuously increments the number in the Y register by 1 accumulates approximately 650 counts per second. The speed of the calculator is likely an artifact of the machine originally being designed as a computer, with logic operating at higher speeds than most calculators. The machine has two indicator lights on the keyboard panel for warning of error conditions; the "MACHINE ERROR" and "PROGRAM ERROR" lamps. The MACHINE ERROR indicator lights to indicate a fault with the machine, or when the cassette tape device gets an I/O error. Pressing the [PRIME] key will clear the MACHINE ERROR indicator (hopefully, assuming there is not some kind of hardware problem) and return the machine to normal operation. The PROGRAM ERROR lamp indicates an error in the running program, such as a branch with destination that can't be found, or running off the end of program memory. Pressing the [PRIME] key will clear the machine, and the PROGRAM ERROR condition.
Date Stamp on 720C Circuit Board
This particular 720C appears to have been manufactured sometime in the mid-1972 timeframe. The final Q/A inspection sticker lists a date of August 16, 1972. Along with the date stamps provided on the circuit boards and other components, the date codes on the IC's seem to match up well with the build timeframe of this machine.