+Home | Museum | Wanted | Specs | Previous | Next |
Wang 720C Advanced Programming Calculator
Updated 5/14/2021
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 Ms. 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 Ms. Harrison and her family for making Mr. Harrison's prized equipment a lasting exhibit in the Old Calculator 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 lower-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 chip sets condensed the entire functionality of a scientific calculator down to a handful of IC packages. Handheld calculators were becoming prolific, and margins in the calculator business were dwindling due to low-cost calculators manufactured in Japan. 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 chip manufacturers. He rightly decided to gradually phase Wang Labs out of the calculator business, concentrating the engineering resources of the company on developing word processing and small business computer systems. Wang Labs 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, and microcomputers had begun to advance to the point where they were easier to use and program than a programmable calculator like the Wang 720C.
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 of highly enough that they decided to build a machine of their own that at least matched the Wang 700. This machine was called the Electronika D3-28. The Soviet-built calculator is, for all intents and purposes, a functional replica of the Wang 700, but packaged in a much more sleek, refined, compact, and service-friendly cabinet, with a cabinet design reminiscent of HP's competitor to the Wang 700-series, 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 copy of Wang's unique wire rope ROM for microcode storage. Like the Wang 700, the D3-28 is implemented using 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 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, and was much easier to interface to. The microcode ROM of the D3-28 has the same 43-bit microcode word size as the Wang 700, indicating that perhaps the microcode is variant of the microcode for the 700. There are some signs that the microcode, while very similar to that of the Wang 700, 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 larger solid-state memory, may have subtle microcode changes to allow it to address more memory than the 700-series machines.
The Wire Rope 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 very likely that the Soviets got their hands on a Wang 700 and reverse-engineered it, and along the way, incorporated changes necessary to adapt the Wang design to their available 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 key-switches rather than the micro-switch 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). While the keyboard design is different, the layout is very similar, with key groupings the same as Wang's machine, and in many cases, the same functions on each key location. 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. Vaporware refers
to the advertisement of equipment that is not yet in production, or perhaps
isn't even yet in operational prototype form in its worst case. Technology
companies, especially back in those days, would frequently advertise products
as if they were real products, even though they hadn't yet produced any of
the product. This was seen as a way to get customers to pre-order the product,
allowing the sale to be booked even though the product might not be delivered
to the customer until significantly later. In such advertising, mock-ups
or dummy versions of the calculator being offered were used in the photography,
and the specifications stated were best guesses of what may be those
of the production machine in its final form.
On top of the high-end calculator marketplace with its
300-series calculators, Dr. Wang felt that his company could divert some of
his company's engineering resources to compete in the commercial computer
market. This market was dominated by IBM, along with Wang's neighbor
located in Maynard, Massachusetts, Digital Equipment Corp. (DEC), and a
smattering of other players such as General Electric, RCA, Burroughs, and
Control Data. Dr. Wang wanted his company to build a computer that could
compete head-to-head with IBM's wildly successful IBM System/360, and win.
Dr. Wang wanted a computer that Wang Labs could 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.
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 was visibly shaken. He thanked Bill Hewlett for showing him
HP's latest brainchild, and let the room muttering
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 transistor-based desktop package. Not long after HP
began shipping orders for 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 counter HP's wunderkind.
Dr. Wang went to the folks designing Wang Labs' 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), scaling down of the microcode to control a smaller
calculator-specific register and processing architecture, and the writing of
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 instantly changed his mind about trying to do battle
with IBM, and focus more on saving his calculator 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, along with Koplow, were tasked with the job
of converting the logic design for
the computer into a high-end calculator. Mr. Shu-Kuang (S.K) 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 had critical mass.
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 skill set 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 perseverance 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 nearly half a year.
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 (which was likely a mock-up rather than an operational
unit).
Early Prototype Wang 700-Series Calculator
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 subtraction
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/subtraction
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.
Finally, an entire fourteen months after the initial announcement,
in the February 1970 edition of Wang Laboratories' monthly publication, the Wang Laboratories
"Programmer", an official
announcement article for the Wang 700
was published, not so subtly bragging that the machine was the fastest and most capable
electronic calculator on the market. Wang Labs used over a year of
hyping the 700 to the marketplace as a means to try to assure
current and potential customers that they had a machine that could
counter the HP 9100A/B calculators, and through it would be ready
"soon", that they could place advance orders for the machine
and be first in line to get one when it became a reality.
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 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...by a serious margin.
Pre-Production Wang 700A, Late 1969 By late 1969, Wang Labs have finally
got all of the details of producing what was the Wang 700A calculator sorted
out. Pre-production machines had been built and were going through
various testing processes to assure that they would survive the rigors
of shipping and day-to-day use. The machines where drop-tested and vibration
tested, checked for proper operation at marginal mains power, safety tested
for proper grounding and short-circuit protection, and subjected to extreme
heat, cold, and humidity conditions to assure that the calculators would
survive just about any conditions that they could be subjected to.
These pre-production calculators were never intended for sale to a customer, but
were used as demo machines for trade shows, field sales offices, and customer
loaners if the customer machine required a visit to the shop for repair. As
shown above, a pre-production machine was also used for some somewhat
begging advertising for the
calculator beginning in February of 1970, stating that the machine was in
production and finally, long-standing orders were finally read to be
fulfilled.
Built-in Cassette Tape Drive for Program Storage
Computer systems of the time were generally 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). 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) 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, conveniently 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
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 superseded 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 retrofitable, 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 key caps to help the user quickly find the function
they are looking for. Clear key caps are generally related to numeric entry
and control functions, blue key caps indicate higher-level math functions
and other specialized functions, rose-colored key caps are related to data
storage and retrieval, and smoke-gray key caps 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 key caps 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 key cap. The snap-on cover is made of
clear (or tinted) plastic to allow the key cap nomenclature to show through.
The Keyboard Circuit Board, with myriad Micro-switches 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 micro switch, such that when the switch
is pressed, the plunger of the micro switch is depressed, activating the switch.
The spring pressure of the micro switch pushes the key cap 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
lives on group of twelve plug-in circuit boards containing 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 enameled magnet wire are strung across the board, stretching
from the diode selection logic up through wire guides, and into
square-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, co-founder of Wang Laboratories, and, at the time, Wang
Labs' 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 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 (3,639,965) 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
toroidal-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 88,064 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 molded 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 [2] 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
A Later Wang Cassette Tape 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 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 time frame. 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 time frame of this machine.
Image Courtesy Bill Needler
Image Courtesy Bill Needler
Note Location of Wang Logo(Above Cassette Door), and missing "Wang 700 Advanced Programming Calculator" on the Cassette Door
Note no fan on the back of the 700-Series Cabinet
Image Courtesy Bill Needler
Photo from February, 1970 Wang Laboratories "Programmer" Periodical
Image Courtesy Bill Needler
Item Courtesy of the Thomas J. Harrison Family
Item Courtesy of the Thomas J. Harrison Family