Remington-Rand EDC-III Desktop Calculator
The Remington-Rand EDC-III is the last model of Remington-Rand's first-generation electronic calculators. After these first-generation machines, the company quit producing their own electronic calculators, and marketed calculators built by other vendors, primarily Casio, under the Sperry/Remington-Rand brand. The first-generation machines, as currently known, consisted of the EDC-1, EDC-1D, EDC-III, and EDC-IIIA. There is no known reference to any EDC-II-series calculators. If anyone out there knows of anything of the early history of Remginton-Rand's electronic calculators, I would love to hear from you. You can click the EMail box in the navigation bar at the top of the page to reach me.
All of the machines in this first generation shared a similar physical and electronic design, utilizing small-scale integrated circuit logic; a magnetostrictive delay line for working register storage; extremely high build-quality (almost to military specifications); a highly-readable display utilizing Nixie tubes with 5/8ths-inch tall digits; an absolutely fantastic Hall-effect keyboard, and an absolutely annoying tone generator with speaker that emits varying tones depending on the type of error condition encountered.
The EDC-1 was the company's first entry into the fiercely competitive electronic calculator market of the late 1960's. It appears that the EDC-I was introduced sometime in early 1968, with the EDC-1D coming slightly later. The EDC-I was quite a primitive machine for its time, vastly out-classed by many earlier electronic calculators. The EDC-1 operated with integers only -- the machine had no capability for handing anything behind the decimal point. There is no decimal point key on the keyboard! Any calculations requring fractional capabilities would require that the user manually keep track of the decimal point location. A small plastic sliding marker below the display window provided a manual means to assist the user in keeping track of the decimal point. At the time, it was considered more important that the EDC-1 provided double-precision (20 digit) capacity rather than provide capability for calculations on fractional numbers. Clearly, the lack of ability to calculate with fractional numbers ignificantly limited the usability of the EDC-1 in many disciplines. To solve this problem, the EDC-1D was introduced, which provided for various fixed decimal point settings, and the ability to enter and calculate with fractional numbers. Even with its ability to work with fractional numbers, the EDC-1D was still rather weak in capabilities, with only the four basic math functions, and a simple accumulation mode for automatically forming sums of products, but no real memory functionalty. Sometime in the fall of 1968, the EDC-III was introduced. The EDC-III was an attempt to play "catch-up", given the weaknesses of the EDC-1 and EDC-1D. The EDC-III utilized small-scale DTL Integrated Circuit technology; provided two independent accumulators; a dual-mode ten-digit display capable of delivering results up to 20 digits; a sum of products and sum of products/quotients mode; automatic constant for addition and subtraction, as well as switch-enabled constant for multiplication and division. Sometime after the EDC-III was introduced, likely in early 1969, the EDC-IIIA was introduced. However, it is unknown at this time what changes were made to the EDC-IIIA. Given the marketplace, one could hazard a guess that perhaps the EDC-IIIA was a redesign of the EDC-III utilizing newer technology (perhaps replacing the magnetostrictive delay line with MOS shift register IC's) to reduce manufacturing cost in the insanely competitive marketplace of the time.
Inside Views of Remington Rand EDC-III
The EDC-III exhibited here was built in the early part of 1969, based on date codes on various components inside the machine. The earliest parts are date-coded in late '68, with the latest parts with date codes from the 7th week of '69 (early February). While the calculator is rather conventional for its time from a technology standpoint, the construction of the machine is unusual. This has to be one of the most well-built and cleanly-designed that machines I've come across.
Heavy-Gauge Cast-Aluminum Baseplate with Power Supply and Interconnection Wiring (Note Speaker)
The base plate of the machine is a heavy-gauge cast aluminum piece. On the base plate are mounted the various power supply components, along with connectors and hand-wired interconnects for the keyboard and logic assemblies. These connectors are high-quality "Mate-'n-Lock" multi-pin connectors that are color coded to assure proper connectivity. The power supply is a conventional design, using a fairly good-sized transformer, diode rectifiers (which, in this unit, have been replaced at one point in time), and circuitry for providing various regulated and unregulated voltages. Both mains and logic-supply fusing is provided.
Unique Flex-Circuit Backplane Interconnect System used in EDC-III
The logic assembly of the EDC-III is the most interesting design I've ever seen on an early electronic calculator. The set of five circuit boards that make up the logic of the calculator are interconnected by an amazing system of flexible circuit material. The logic package can be removed from the machine by removing five long screws, and unplugging two connectors, and simply lifting the assembly off the base plate. The logic assembly is bound, like a book, at one end, with the flexible circuit backplane, and spring-steel "bindings" that allow the logic assembly to be opened, board at a time, again, like a book. The flexible circuit material appears to be metalized (copper) circuit traces, embedded in a clear plastic mylar film. Such flex-circuits are very common today, but in the late 1960's, this was pretty amazing interconnection technology. It's likely that this flex-circuit technology was a hand-me-down from Sperry-Rand's involvement in military electronics.
The top board (Display) in the circuit board stack
While this design is certainly unique, it seems that it would be extremely difficult from a service perspective to replace a circuit board, as all of the flex-circuit connections would have to be un-soldered and re-soldered to a replacement board. My guess is that individual board replacement was simply avoided by service technicians carrying around entire replacement logic assemblies. In fact, an early advertisement for the EDC-III boasts "we've got a card library that can be changed in your own office in 20 minutes". Given this statment, it's pretty safe to assume that the whole logic "library" would be removed and replaced with a known good unit, and the failed logic assemblies would be returned to the factory, where the the fault(s) could be diagnosed, and defective components replaced without having to disassemble the logic assembly.
The second board
The logic of the EDC-III is made up of small-scale DTL (Diode-Transistor Logic) integrated circuits in dual-inline ceramic packages. There are a total of 228 integrated circuits used in the machine, with chips made mostly by ITT, some Fairchild, and a pair of Philco (Part #PD993359) IC's in the keyboard assembly. This high component count makes the machine one of the higher parts-count "basic four function" calculators in the museum. There are a total of seven different part numbers of IC's (including the above referenced Philco IC's), but it appears that Remington Rand used their own internal (sometimes known as "house") part numbering scheme (of the form 1037xx), that makes positive identification of the functions of the chips difficult without some electronic detective work.
Examples of some of the ICs used in the EDC-III
With a manufacturer's suggested retail price of $970 at introduction, along with the complexity and high quality of the calculator's construction, it seems unlikely that there was much profit to be gained in each sale of the EDC-III. This, unfortunately, was somewhat common in the late '60's, as technology was advancing at such a tremendous rate that machines were typically technologically out of date by the time the design was complete and the machine was ready for manufacture and marketing.
The third board in the card "library"
The majority of the ICs are found in the logic assembly, but a few (10) are used in the keyboard assembly for encoding purposes. Along with the IC's, there is the usual assortment of transistors (mostly for clock generation, indicator lamp drivers, and Nixie tube drivers), diodes, resistors, and capacitors. The circuit boards are built with a high level of quality, using blue fiberglass base material, with copper conductors on both sides of the circuit boards with plated-through holes. The boards are flow-soldered, so all of the bare copper traces have a thin coating of solder covering them.
The fourth board
The end of the circuit boards where the flex circuits create the backplane interconnections have solder-plated rectangular pads that the flex-circuit material connects to. It is my guess that some kind of thermal process was used to bond the flex circuit material to the circuit board, creating both an electrical (via the solder), and mechanical (adhesive) connection.
The bottom board (Delay Line Underneath)
Cast-aluminum stiffeners on both sides each board also provide additional mechanical stability for the flex circuit material backplane, which is sandwiched between the circuit board and the stiffeners. Connections between the circuit boards are made via flexible circuit connections on both sides of each circuit board.
Back side of EDC-III Keyboard Assembly
Along with the ususual design of the logic assembly, the EDC-III uses a rather unique keyboard design. Most all early electronic calculators used one of three different methods for changing the mechanical input from the user's fingers into electronic signals for the logic to process. The most common keyboard design uses magnetically actuated reed switches. With this method, a small magnet is mounted at the end of each keystalk. When a key is depressed, the magnet moves into the vicinity of a magnetic reed switch, which closes under the influence of the magnetic field. The other method involves using simple "leaf" switches, where the key stalk presses one metal movable "leaf" contact into another fixed leaf, creating an electrical connection. Less-common, but still frequently used are where the keystalk directly or indirectly operates micro-switches, a snap-action switch assembly that is sealed. Wang and Friden used these type of swithes in many of their calculator keyboards.
Back side of Keyboard Assembly minus electronics. Note high-quality keystalk construction
The EDC-III uses a totally different design, using instead what is called a "Hall-effect" switch. Like reed-switch keyboards, each keystalk has a small magnet in it.
Detail of Keystalk Construction
A circuit board mounted beneath the keyboard contains small black plastic bobbins with a few turns of magnet wire coiled around each one. There are as many of these bobbin assemblies as there are keys on the keyboard. Each keystalk passes through the center hole of the bobbin. When a key is depressed, the magnet in the keystalk moves downward through the center of the bobbin, and the magnetic field created by the magnet in the keystalk induces a change in a small current flowing through the coil, which is detected and amplified by a transistor (one for each key), and sent off to other circuitry for encoding of the keypress for use by the logic.
Keyboard Coil Circuit Board
The keyboard assembly consists of a heavy-gauge sheetmetal plate that serves as a backbone that contains the keyswitches, and two circuit boards stacked one atop another below the keys. The Hall-effect board contains the coil windings and transistor amplifiers, and is located underneath the keyboard plate, with the encoding board mounted beneath the Hall-effect board.
Keyboard Encoding Circuit Board
The encoding circuit board turns the keypresses into binary codes that are recognized by the rest of the electronics, as well as providing keyboard error detection circuitry that detects the accidental depression of more than one key at a time. The Hall-effect design is electronically complex, but also electronically clean, meaning that there's no electrical "bounce" caused by mechanical switching action. It also makes for a smooth and quiet keyboard from an operator's point of view. This design features no moving parts (other than the key itself), making for a very reliable and long-lived keyboard. The keyboard on the exhibited EDC-III operates as smoothly and positively now as it did when new over 30 years ago. While this design is more costly to produce than other more conventional keyboard designs, it is clear that the designers of this calculator were more concerned about quality and reliability than bottom line cost.
The Delay Line under bottom board of logic stack(in aluminum housing)
The magnetostrictive delay line used to store the working registers of the calculator is located at the bottom of the stack of five circuit boards that make up the logic "library". There are no indications that I can find on the delay line that give a clue to its delay time or manufacturer. It is slightly unusual that the input and output each have three connections. Usually there are only two connections. Why this is remains unclear. The delay line is sealed, and all of the fasteners holding it together are staked to prevent non-destructive disassembly. It is clear that the delay line was not designed for serviceability. In the event of a delay line failure, the entire logic assembly was likely replaced, with a new delay line included as part of the package.
The master clock rate of the calculator is 2.0 MHz, as indicated by a crystal located on the middle circuit board in the stack. It is unlikely that this basic clock frequency is used "as-is", but is surely divided down by a factor of some multiple of two for an operational rate of something around 500KHz to 250KHz, more in line with a conservative nominal operational speed for the DTL integrated circuits and the delay line.
Keyboard Layout on EDC-III
From an operational perspective, the EDC-III is somewhet unique. The machine appears to use five main registers; a display register (for entering numbers into and displaying results), two accumulators for addition/subtraction (with the accumulator to be used selected by a slide switch having two positions marked "I" and "II"), a register for storing the first number of a multiplication/division calculation, and a counter register used for multiplication and division calculations. Addition and subtraction work "adding machine" style, with the [+] key adding the number in the display register to the selected accumulator (see below), and the red [-] key subtracting the number in the display from the selected accumulator. Upon completion of either addition or subtraction, the content of the selected accumulator is automatically copied into the display register. Multiplication and division are a little unusual, with a special key for entering the the first number of a multiplication or division. After the first number is entered into the machine, the [X ENT ÷] key is pressed, which enters the multiplicand or dividend, then the second number of the operation is entered, followed by the [X =] or [÷ =] key, which calculates the result, and copies it into the selected accumulator and display register. The red [X -] key is used to cause the second number entered in a multiplication problem to be made negative before the multiplication occurs, calculating the result as ([Multiplier] X [Multiplicand X -1]), and placing the result in the selected accumulator, and copied to the display register.
There are three different keys that can be used to clear various sections of the calculator. The light-blue [ALL CLR] key clears all of the working registers of the calculator, including the two accumulator registers. The [C] key clears the display register, and is used for correcting entry errors. The light-blue [TOT CLR] key recalls the content of accumulator I or II (as selected by a keyboard slide-switch) to the display register, then clears the selected accumulator.
As mentioned before, the EDC-III has two accumulators. When math operations are performed, the particular accumulator to be operated upon is selected by the left-most slide switch at the lower left-hand corner of the keyboard. The accumulators are independent, which allows some interesting flexibility for performing unusual math operations. Along with the slide switch that selects which accumulator the machine is to use, there are two more slide switches. The middle slide switch (labeled "X" as the upper, and "N" as the lower positions) causes the result of multiplication operations to be added to the selected accumulator when in the "X" position. When in the "N" position, the result of multiplication over-writes the content of the selected accumulator. Lastly, the right-most slide switch, labeled "GT" and "N" controls the accumulation of grand totals for multiplication and division operations. When in the "GT" (upper) position, results of multilication and division operations are automatically added to the selected accumulator. When in the "N" (lower) position, the result of multiplication or division calculations overwrites the content of the selected accumulator.
There are three other keys on the keyboard that affect the accumulators. The [TRAN] key copies the current content of the selected accumulator to the display register without disturbing the value in the selected accumulator. The [STOR IN] key copies the content of the display register into accumulator II, overwriting the content of the accumulator. The [STOR USE] key copies the content of accumulator II into the display register, leaving the content of accumulator II undisturbed.
The EDC-III operates with fixed decimal. The decimal point location is selected by a large thumbwheel switch located at the left-top of the keyboard panel. The switch provides for decimal point selections from zero to ten digits behind the decimal point. Entering any more digits behind the decimal point than selected by this switch will cause an input overflow. When a number is entered into the display register and an operation key is pressed, the number in the display is automatically corrected to present the correct number of digits behind the decimal point, with trailing zeroes filling unused positions.
Addition and subtraction always operate in constant mode. The last number entered before pressing the [+] or [-] keys is remembered as the constant. Pressing the [+] or [-] key again without entering another number causes the operation to occur with the last-entered number. For example, with the decimal point selector set at 4, pressing , followed by the [+] key will result in "000001.0000" in the display. Pressing the [+] key again will display "000002.0000". Pressing [-] will result in "000001.0000", etc. Multiplication and division do not normally operrate with a constant. However, a rocker switch on the keyboard labeled "CON X ÷" allows the first number entered in a multiplication or division calculation to be retained as a constant. This switch is automatically returned to the "off" mode when the [ALL CLR] key is pressed, via a mechanical linkage between the [ALL CLR] key and the constant mode rocker switch.
Display views with "DISP" rocker switch in "low-order"(top) and "high-order"(bottom) digit selections. Display reads out 5465076.127450
An extremely unusual feature of the EDC-III is the fact that it can perform double-precision arithmetic, essentially allowing entry and results of numbers up to twenty digits. An indicator panel (which includes a "-" nomenclature for sign infication, "OVERFLOW" for overflow location, and a leftward-facing arrow for indicating that additional digits can be viewed by operating the "DISP" rocker switch) is located to the left of the display panel. When a number is in the display register that contains greater than ten digits, the left arrow indicator lights to indicate that the display currently contains the lowest order ten digits of a larger result. Pressing the "DISP" rocker switch swaps the display so that the upper-most ten digits are displayed. For example, if the calculator is set for 6 digits behind the decimal, and the calculation "9301.522549 + 100000 X 50" is performed, the display will show "<-5076.127450", with the "<-" (arrow) indicating that there are more significant digits in the result. Pressing the uuper part of the "DISP" rocker swtich will display "<-0000000546", indicating the upper-most digits of the result. In total, the result is "5465076.127450", a total of 13 digits of result from a calculator with a 10-digit display. The "DISP" rocker switch shows the least-order digits of the 20 digit display register when the lower half of the switch is depressed, and the upper-order digits when the upper half of the swtich is depressed. Note that in the calculation above, that "100000" is larger than can be entered with the decimal point set at six digits behind the decimal. Because the display register actually contains twenty digits, this entry is possible. The display register can provide up to twenty digits of display capacity, however, the actual display capacity is determined by the setting of the decimal point location. The actual number of digits of capacity of the machine is determined by subtracting the number of digits set behind the decimal from 20. For example, if the machine is set for 7 digits behind the decimal, the machine is only capable of developing results up to 13 digits in length. At this decimal point setting, any result in excess of 13 digits will cause an overflow condition.
As mentioned early-on in this exhibit, the EDC-III has a very annoying
feature. When an error condition exists (any of calculation or input
overflow, depression of more than one key at a time, or division by zero)
a small speaker located on the baseplate of the calculator emits a rather
loud tone (whose frequency varies depending on the error condition type, a
higher-frequency tone for overflow, and a lower-frequency tone for
input error) that can only be silenced by pressing the [ALL CLR] or [C] key.
There is no means provided to disable this tone, nor to change its volume
level. This particular "feature" had to be extremely annoying in office
environments, where the whining of the speaker would announce to the rest
of the office that the operator had made an error. One can not fathom the
reason why the engineers that designed this machine with such a quiet and
smooth (albeit complex) keyboard thought that this noisy error indication
was useful in an office environment.
It quickly became very annoying to the author while putting the machine
through its paces. This "feature" most assuredly had to have driven office
workers crazy, and it's my guess that many service calls were placed by
customers to have the speaker disabled.
The EDC-III does not provide power-on reset. Most of the time the display shows all zeroes on power-up, occasionally there can be garbage (Nixies with multiple digits lit at once) in the display register and accumulators which can lead to incorrect results unless the [ALL CLR] key is pressed before beginning to use the calculator after power on. The operator's manual states that it is very important to press [ALL CLR] before using the calculator after powering on.
The calclation speed of the EDC-III is not particularly notable. It is a bit slower than many of the calculators of similar timeframe. This is likely due to conservative design practices, where clocking speeds are kept well-below the maximum rated speed of the IC's to minimize the chance of any kind of IC tolerance variations causing improper operation. During calculation, the Nixie tube display is blanked, so there's no drama in the calculation process. The "<-" indicator lights and flickers a bit during subtraction and division, probably an artifact of the way these calculations are processed. This indication does not happen during addition or multiplication. Addition and subtraction complete in perhaps 50 milliseconds or so. Subtraction seems slightly slower than addition. Multiplication takes at the most 1/2 of a second (for 999999999 X 999999999, note only nine 9's...using more than 9 nines results in an incorrect answer). Division can take significantly longer, with 18 nines divided by 1 taking approximately 1.2 seconds to complete. Trying 19 nines or 20 nines (with the decimal point location set at 0) yields in an incorrect result. The fact that the maximum capacity of the machine can't be utilized is likely due to the high-order digit or two in the various registers of the machine being used for "housekeeping" processes during multiplication and division calculations. Activating either or both of the two accumulation modes makes no appreciable difference in the timing of multiplication or division calculations.