Toshiba "Toscal" BC-1411 Desktop Calculator

This fantastic machine comes to the Old Calculator Museum through the courtesy of Mr. George Kania, who works in Computer and Network Support at the University of Newcastle upon Tyne, in Northeast England. Mr. Kania contacted the Old Calculator Museum via EMail indicating that he had this machine, and was wondering if the museum be interested in purchasing it. Due to the limited funding the museum has, it was not possible to offer a reasonable sum for such an unusual and unique calculator in what appeared to be stunningly-preserved condition, but Mr. Kania was very gracious and agreed to an affordable price. After a number of difficulties getting payment transferred to the UK, the calculator finally started its journey from the UK to the museum in Beavercreek, Oregon, USA.

Mr. Kania had sent high-quality photos of his machine, and it looked from the photos that the machine was virtually like new. It is unusual for a calculator of this vintage to be in such pristine condition. The museum curator was looking forward to the machine arriving, but apprehensive as shipping such a vintage machine from the UK to US is fraught with risk. The thought of this machine being damaged in transit was the stuff of nightmares.

The trip took eight days. The box definitely looked like it had been on an adventure, with a good-sized divot in one corner and part of the top was somewhat mashed in on one side. Fortunately, the packing job was exceptional, with double-boxing and lots of padding. The interior box showed no signs of damage - the outer box absorbed all of the shipping abuse. After extracting the machine from a thick layer of bubble wrap, I was happy to find that there were no broken Nixie tubes or damaged keyboard keys...the most likely items to receive damage. In fact, everything looked really good, which was quite a relief.

Very early advertisement for the Toshiba BC-1411

Leading up to the calculator showing up, research into Toshiba's history in the calculator business was performed. The search for information about Toshiba's calculator business proved to be very difficult. It seems the story of Toshiba's calculator business, and much of the information about its early electronic calculators are lost to the sands of time. Remarkably, Toshiba's online history site doesn't even mention their foray into the calculator business. A query was made through Toshiba's corporate web site asking if anyone had any information on Toshiba's calculator business. Promptly a very kind response was received indicating that there simply was no information left. Toshiba left the calculator business in the early 1980's, and when they did, nothing was kept. The people that worked in the calculator division all eventually left, retired or passed away, and the kind man that answered the query said that there just wasn't anything to be found. So, the information presented within this exhibit is a patchwork assembly of data derived from old advertising and business machine trade publications.

Based on date codes on parts found within the machine, it was likely manufactured in the late 1967 time frame. From what little information is available on Toshiba's calculator history, the BC-1411 began production in November, 1966. The BC-1411 is the second electronic calculator that Toshiba marketed, with the first electronic calculator called the BC-1001, introduced in December of 1965. The BC-1001 was a 10-digit, four function, fixed decimal, Nixie tube display machine, likely using very similar technology to the BC-1411. The BC-1001 may only have been sold in Japan, as at the time, Tokyo Shibaura Electric Co. was working on establishing a presence for marketing their calculators in the US.

Before powering up any old machine like this, it is imperative to do a complete visual inspection to make sure that nothing shifted in shipment; for example, that all of the circuit boards are properly seated in their sockets, and that all connectors and wiring is secure, and that there is nothing foreign (such as staples or paper clips, which commonly end up inside old office machines) floating around inside the machine that could cause a short circuit. All fuses are also tested, and the power switch and transformer are checked for shorts. The power cord is also carefully checked over to assure that it is safe. This calculator proved to be in very clean and incredibly good condition inside and out. There appeared to be absolutely no issues that occurred related to its being shipped. This is a testimony to the well-built construction of the machine, and the quality of the packing of the machine for its journey to the museum.

The exhibited calculator was originally sold in the UK, and thus uses 240 Volt European line power. For initial testing purposes, the machine was connected up to a conversion transformer that converts the US 120V line power into 240V. The conversion transformer was plugged into a Variac, so the line power could be slowly ramped up while carefully observing the machine, monitoring the power supply voltages with an oscilloscope and digital voltmeters. As the power was very slowly brought up over the course of about 90 minutes, everything looked good.

The reason for such a time-consuming power-up process is that the power supply filter capacitors in old electronics like this can become defective when sitting without power for long periods of time. The effect that occurs creates resistance in the capacitor. The increased resistance diminishes the AC filtering capabilities of the capacitor, and also causes the inside of the capacitor to heat up when current is applied. The internal buildup of heat has the effect of causing more electro-chemical problems in the capacitor, which can cascade in such a way that catastrophic failure of the capacitor can occur, ranging from the capacitor venting electrolyte to a pressure build-up inside the capacitor that results in a small explosion and potentially fire. Such failures can cause serious damage to other components in the calculator. The procedure allows the capacitors a soft start that will expose any potential problems before the voltages get to a point where severe damage can occur. This is why it is very important to avoid the temptation to just plug-in a newly found old calculator.

Side profile and top quarter view of the stylish BC-1411

Fortunately in this case, the capacitors were quite healthy, and the power supply voltages were clean with very little ripple. At around 100V AC on the Variac (which would be about 200V at the calculator) the Nixie tubes started flickering to life. The Variac was left at 90% of rated voltage for about 10 minutes, allowing the circuitry to warm up and for all of the Nixie tubes to light up with a zero. Once all the tubes were lit, The [C] (Clear) key was pressed, and the decimal point lit up at the right end up the display. Then the [1] digit key was pressed, and low and behold, the display read "00000000000001.". Following that key press with the digit [2], resulted in "00000000000012.". This was a very good sign. The Variac was then slowly turned up to 100%, the [C] key was again pressed, and a general process was begun to test the various functions of the machine. Delightfully, the machine proved to be fully operational through the initial testing.

After testing the machine for a while, the only fault that was found was that the overflow indicator lamp appeared not to be operating properly. Operations that overflowed the capacity of the machine wouldn't cause the lamp to light. In the big scheme of things, this was a very minor issue. As it turned out, the lamp was simply insecure in its socket. Re-seating the lamp in its socket solved the problem.

The unusual capacitive dynamic RAM (Random Access Memory) board used for working register storage in the BC-1411
Click on board image for a larger view and description

The next step after performing the testing and finding the machine to be basically sound is to completely disassemble the machine, cleaning it internally and externally(though in this particular case, little was needed), and taking the various photographs to be used in the exhibit. Also during this time, a more thorough visual inspection is done of each circuit board and subassembly to look for any lurking problems. Through this process, it was discovered that this machine was quite unique compared to other calculators of this vintage. The revelation was that the machine did not appear to use any of the common means of storage for the required working registers of the calculator. There was no magnetic core memory array, no magnetostrictive delay line, no chains of flip flops or ring-counter configurations. It was clear that that some other mechanism had to be used for storing the working registers of this machine. After inspecting all of the circuit boards, it was surprising to find that the BC-1411 uses this unique method for working register storage; essentially a mid-1960's implementation of memory using essentially the same principle used for the main memory in today's personal computers, only on a much larger size scale, and vastly smaller capacity!

The other seven circuit boards from the BC-1411
Click on board image for a larger view and description

The dynamic memory chips in today's computers rely on tiny capacitors that store a minute electrical charge. If a charge is stored in a capacitor, it represents a logic 1. If no charge is in a capacitor, a logic 0 is represented. The problem with using capacitors as a storage medium is that the charge in a capacitor naturally bleeds off over a relatively short time, resulting in a memory that forgets that a 1 is stored in it, with the charge eventually bleeding down to the 0 level. The solution is to periodically read and re-write the charge in each capacitor that contains a 1 such that the charge is always maintained. This process is called a refresh cycle. In today's computers, the dynamic RAM (DRAM) chips contain millions of microscopic capacitors and associated circuitry integrated onto a tiny chip of silicon, and the refreshing of the memory is controlled by the sophisticated memory controller in the computer. In the Toshiba BC-1411, the technology takes the form of discrete capacitors arranged in a 15 X 12 array (for a total storage capacity of a whopping 180 bits) that stores the 1's and 0's that make up the three working registers of the calculator. Clever architecture assures that each of the 180 memory cells is refreshed frequently enough that the capacitors form a reliable memory system for the calculator.

The BC-1411 without cabinet, showing the robust construction of the machine

In the process of the detailed checkout, it was found that the build quality of the BC-1411 is very high. It is built more along the lines of German calculators like the early Olympia machines (also OEM'd to Monroe in the US, such as the Monroe 740) rather than many other Japanese manufacturer's generally less robust construction. The top and bottom halves of the cabinet are both made from heavy-gauge sheet metal. There is no plastic in the chassis or cabinet of this calculator. The upper part of the cabinet is held in place by two captive screws on the back panel, and two latch devices near the front of the machine. Simply loosening the captive screws, and flipping the latch handles that protrude through the bottom of the chassis allows the top cabinet to easily be lifted off. The calculator's entire chassis is made of heavy gauge sheet metal. The machine has a very modern look to it, with a separate nacelle making up the display area and a nicely sloped keyboard bezel providing a smooth look to the machine. All of the metal work and finishes are of high quality. The machine would have made an impressive and stylish addition to any office of its day.

The card cage and backplane wiring of the BC-1411. Note extensive card guides.

The logic backplane consists of a thick sheet metal baseplate, to which wire wrap-tail edge-connector sockets are secured by screws. The backplane is wired with very neatly arranged individual wire-wrapped interconnects. A set of nylon edge card guides on the sides of the card cage, along with board separators along the bottom of the card cage assure that the cards are held firmly in place, and that the boards don't interfere with each other. To add to this already extensive support, two top straps with medium-density rubber blocks fit in place across the top of the card cage, holding the tops of the circuit boards in position as well as providing some shock isolation. The circuit boards are made of standard copper-clad phenolic, and appear to have been carefully processed, as the edges of traces are very clean. The edge connector fingers are gold-plated, with an unusually thick layer of gold.

Close view of backplane wire-wrap connections

The circuit boards are nicely laid out, with a functional units grouped together. There are even some etched notations on the board indicating the function of groups of circuitry, which is thoughtful, though uncommon. The boards have traces on both sides, utilizing plated-through holes to interconnect both sides of the board. There are a total of eight circuit boards that plug into the backplane. Each board, which is approximately 12-1/2" X 7-1/2" in size, has two sets of edge connector fingers, each with 56 fingers, 28 front, and 28 back. The sockets jumper both sides of the board fingers together, so each connector contains only 28 connections, for a total of 56 connections per circuit board. The edge connector sockets also have gold-plated contacts for long life and superior conductivity.

The display subassembly of the Toshiba BC-1411

The Nixie tube-based display assembly is also made from heavy gauge stamped sheet metal, forming a substantial cage surrounding the fourteen Nixie tubes, providing shock isolation and secure positioning for the tubes. The Nixie tubes are Japanese-made CD-66 tubes, with 5/8"-inch tall digits, and a right-hand decimal point. Date codes on the Nixie tubes indicate mid-1966 as the date of manufacture.

Back side of one of the Nixie tubes, showing CD-66 part number, and 6H date code

The Nixie tubes are soldered onto a circuit board the interconnects them, which has a cable bundle attached that terminates in a military-style multi-pin connector that plugs into a mating connection on the bulkhead between the card cage and the keyboard/display area. The display is multiplexed, meaning that the digits are sequentially scanned a digit at a time at such a speed that the display appears continuous. This type of display, versus the direct per-digit drive of similar-vintage calculators such as the Sharp Compet 20 and Canon 161, is much more cost-effective, and very well fits the bit-serial nature of the calculator's architecture. There is no leading or trailing zero suppression in the display.

The Keyboard subassembly

The keyboard assembly of the BC-1411 is also structurally very sound, again made from a skeleton of heavy-gauge sheet metal. The key switches have very high-quality plastic key caps, with molded in nomenclature to prevent wearing off of the legends with extended use. The key switches are a rather complicated assembly that utilize magnetically activated reed switches to provide the switching action. These type of switches provide a very clean switching signal, meaning that less complex circuitry is required to de-bounce the signal. Leaf-contact and micro switches generally require more de-bouncing circuitry as there is more mechanical "noise" introduced into the switching signal. The keyboard operates very smoothly, with moderate return spring pressure making for a positive key press, but without too much back pressure.

Key switch construction detail

Most of the keyboard keys are encoded into a unique bit pattern by a diode matrix. Some of the keys aren't encoded, and connect directly to the logic, such as the accumulation mode key switch. Push-on/push-off keys incorporate the latching mechanism into the key switch module. The keyboard circuitry or mechanics do not prevent simultaneous depression of keys, which can lead to entry errors. Accidentally pressing the [5] and [6] keys together at the same time results in a "7" being entered. Likewise, pressing the [7] and [8] keys at the same time results in a display with a 7 and a 9 simultaneously lit in the same Nixie tube. The main power switch, located at the left end of the keyboard panel, is a hefty push-on/push-off switch with a red key cap. The switch makes a satisfying "thunk" when activated. The keyboard assembly connects to the logic backplane via a military-style multi-pin connector.

Power Supply Topside

The power supply, located beneath the keyboard assembly, is almost computer-grade, with a large power transformer, followed up by large power-diode rectifiers to convert the transformers stepped-down AC to DC. A large computer-grade filter capacitor, combined with a number of smaller filter capacitors mounted on a terminal board are used to smooth the switching glitches of the diodes, followed by transistor-based voltage regulation. Both the primary and secondary voltages of the power supply are fused for safety. The supply delivers +10 and -10V DC to drive the logic, and 180V DC for driving the Nixie tubes. The +10/-10V supplies are adjustable by potentiometers to adjust for component drift. Fortunately, when checked, the power supply voltages were dead-on. The power supply connects to the backplane wiring via a multi-pin bulkhead connector.

Power Supply bottom side showing additional filter capacitors and potentiometers for setting logic supply voltages

The machine does have a power-on reset circuit that generally results in the machine being cleared and ready for option when first powered up. Once in a while, this power-up clear doesn't function properly, and the display may come up with random digits. Pressing [C] followed by the [M] key will clear the machine and the memory register, just to make sure everything is cleared out. I don't know if this behavior is normal for the machine, or is the result of perhaps some components that have drifted out of specification rendering the power-on-clear intermittent.

Bulkhead connectors for providing backplane connections for the keyboard, display, and power supply assemblies

All combined, the hefty construction of the machine adds up to a pretty large-sized device that consumes quite a bit of desktop space, and is quite heavy. The machine takes up nearly 2 square feet of desk space, and weighs in at a healthy 37-1/2 pounds. Dimensions are 16-inches wide, 18-inches deep, and 9-inches high. The machine's logic contains 286 transistors, all Toshiba-made. The majority of the transistors are Germanium PNP 2SA50's, with the only exception being the Nixie tube display driver transistors.

BC-1411 keyboard layout. Note decimal point selection wheel at left, with jewel for overflow lamp above it

Functionally, the Toscal BC-1411 is quite conventional. It is a four-register machine, with three registers stored in the capacitive dynamic memory, and the fourth register made of a series of flip flops organized as a shift register. The shift-register makes up the entry/display register of the machine, into which data from the keyboard is entered, and from which the display is generated. The capacitor storage holds the memory accumulator, a secondary register for storing operands and as a counter for generating products and quotients, and the main accumulator register. The BC-1411 uses a hard-wired state-machine driven control system which was the common means for control of electronic calculators designed during this time period. With state machines, a series of counters and hard-wired gating directs the machine through the various states required to perform its functions. The master clock frequency of the calculator is nominally 40KHz (40,000 cycles per second), which is divided down by the various counter stages to generate the major and minor states which govern the cycling of the machine. In the exhibited calculator, the master clock frequency was measured at 40.6 KHz, slightly faster than the nominal frequency. The calculator's architecture is bit-serial, meaning that every operation occurs on two bits at a time. The adder is a single-bit full adder with a carry flip flop to store carries until the next set of bits are presented. A separate carry flip-flop exists that has logic to detect when a digit-level carry needs to be propagated to the next digit. Through use of clever design, there is no separate cycle for refreshing the capacitive memory. Since the calculator is always cycling through the digits of the display register a bit at a time in order to generate the display on the Nixie tubes with each bit being read out and re-written, to refresh the rest of the bits of the other three registers, additional read/re-write cycles are inserted between the accesses of the display register. By this method, all of the bits in the capacitor memory are refreshed as a normal part of the display cycle of the calculator. Since the display is constantly being generated, even when the machine is performing math operations, the memory is still continuously refreshed.

From a user perspective, the BC-1411 is characteristic of other electronic calculators of the time. It utilizes arithmetic logic for addition and subtraction, and algebraic logic for multiplication and division, utilizing a combination [+=] key for performing addition as well as terminating multiplication and division operations. The machine uses fixed decimal point logic, with the decimal point position set by a thumb wheel on the right side of the keyboard panel. Decimal point settings are available for 0, 3, or 6 digits behind the decimal. It seems odd that there's not a setting for two digits behind the decimal (for dollars and cents calculations), but apparently Toshiba engineers did not consider this important. The additional circuitry to provide two digits behind the decimal point would have been trivial. Changing the setting of the decimal point position while performing calculations can cause the machine some confusion. If more digits are input behind the decimal than are selected, the machine simply discards the excessive input, with no overflow indication. Inputting numbers beyond the capacity of the machine does not result in the overflow indicator lighting. Inputting more than 14 digits simply results in the data input starting over at the right end of the display, with newly entered digits replacing the previous content of each digit position as successive digits are entered. The [C] key clears the entire machine except for the memory register. The [KC] (Keyboard Clear) key zeroes out the entry/display register, generally used for correcting erroneous input.

The Toshiba TOSCAL Identification Tag, one on the keyboard panel, another on the back of the display nacelle.

The BC-1411 has a single memory register which can be used for storing and recalling a number, or as an accumulator to sum the results of multiplication and division operations. The [M] key copies the content of the display into the memory register. The [AM] key is a push-on/push-off key that puts the memory into accumulation mode when locked down. In this mode, the results of multiplication and division operations are automatically added to the content of the memory register. The [R] key copies the content of the memory register to the display. There is no direct way to clear the memory register. One must first clear the display, then use the [M] key to deposit zero into the display register to clear the previous content. The BC-1411 offers a mode for performing percentage calculations. The [%] key is a push-on/push-off key that, when locked down, automatically divides the second number of multiplication and division operations by 100(by shifting the number two digits to the right). For example, with the [%] key locked down, performing 100 [X] 5, then pressing the [+=] key results in "00000000005.000" in the display with the decimal point set at 3 digits. The effective calculation in this case is 100 X 0.05. The BC-1411 provides for chain multiplication and division, but does not provide any means for performing constant multiplication and division.

The rear panel of the TOSCAL BC-1411. Two captive screws are removed to take off the cabinet.

The BC-1411 does not handle negative numbers, which is quite surprising for an otherwise nicely-featured instrument. Negative numbers are represented in tens complement form, e.g., -1 is represented by "99999999999999.". This limitation would have made the machine somewhat less useful than some of its contemporaries in financial applications where the ability to generate true negative results is important. Like many machines of the era, the BC-1411 has some glitches. Division suffers some problems with large numbers, with some calculations that would be expected to deliver correct results giving incorrect results with no overflow indication, and in some cases, results that have invalid digit (the calculator uses four-bit Binary-Coded Decimal (BCD) to represent the digits) codes in them that causes the digit decoding logic to generate invalid states that can cause multiple digits to be lit within a single tube at the same time. Entering multipliers that are too large (for example, entering 11111111111111 with the decimal point location set at 3, then pressing the [X] key) will cause the machine to lock up with garbage in the display, requiring a press of the [C] or [CK] key to unlock the logic. In general, the machine is pretty good about catching overflow conditions when operated within its boundaries, but when multiplication and division calculations are performed that stretch the capacity limits of the machine as mentioned above, frequently overflows are not detected. In normal calculating, causing an overflow condition lights the overflow indicator (located above the decimal point selection wheel). Overflows do not inhibit further calculation, generating results of limited use. The user must be vigilant about keeping watch on the overflow indicator lest incorrect results creep into calculations due to un-noticed overflow conditions. Division by zero puts the machine directly to the task of calculating the incalculable, with all of the decimal points flickering across the display, due to the machine making a futile attempt to shift the (non-existent) dividend to the high-end of the register before beginning the division process. The only way to stop the this confusion is to press the [C] key to reset the machine, or power-cycle it to return it to normal operation.

The Model/Serial Number Tag

With a master clock frequency of only 40KHz, the machine can't be expected to be a speed demon. Clearly the designers of the machine wanted a machine that was extremely reliable, so they didn't push the speed of the machine, rather focusing on running all of the circuitry at comfortable levels. I also suspect that the ~40KHz master clock frequency was convenient once divided down to the full-cycle rate of the machine in terms of providing good timing for the in-built refresh cycle of all of the bits in the capacitive memory. Too fast a refresh rate may not allow sufficient time for a capacitor to fully charge, and too slow a refresh rate could allow enough time between refresh cycles such that the charge drains off, causing the memory to "forget". It's much like Goldilocks' porridge. Things have to be "just right" with the timing of the memory refresh cycles.

The conservative design of the circuitry in the calculator is probably a good part of the reason why this machine is still fully functional today; all of the components were designed to run well beneath their rated specifications, which is one of the more common methods of designing circuitry for long-life. Running components at the edge of their specifications generally results in a shorter service life due to inevitable aging of the components that changes their specifications over time.

Despite the unusually slow clock rate of the BC-1411, the calculator still performs math operations orders of magnitude faster than the fastest electromechanical calculators it was developed to replace. And, it does so without all of the noise made by electromechanical calculators. Addition and subtraction yield results somewhere between 10 and 20 milliseconds - enough to where a flicker of the display is barely visible as the operation proceeds. Multiplication and division can take significantly more time. The conventional "all 9's" times 1 calculation takes approximately 1 second to perform. Thirteen 9's divided by 1 takes a little over a second. Trying the division with fourteen 9's results in an incorrect answer, but takes approximately 0.2 seconds longer than the thirteen digit operation. Performing these operations with the [AM] (memory accumulate) mode turned on doesn't seem to add any appreciable delay to the generation of results. It appears that the accumulation into the memory register is part of the normal timing of the multiply/divide state machines, and rather than skipping over this operation when the [AM] key is not depressed, zero is simply added to the particular digit of the memory register as part of the normal process of generating the result of multiply and divide operations. When the [AM] key is depressed, the resulting digit is added to the memory register and any carry to the next digit is noted to be added in during the next higher digit's cycle. Embedding the [AM] logic into the normal cycles for multiplication and division means that the operations take the same amount of time whether the [AM] mode is on or off.

During calculations, the Nixie tubes are left active, and due to the relatively slow rate of calculation, it is possible to observe some of the shifting and counting operations occurring during more complex multiplication and division operations.