Casio FX-1 Scientific Calculator
The Casio FX-1 was the first true scientific calculator made by Casio, and, if not the first, one of the very earliest LSI IC-based scientific calculators. The FX-1 was introduced in late 1971, with this particular model being built in early 1972. It was purchased by its original owner from W.T. May Company in Jackson, Mississippi on March 29, 1972. This machine is in wonderful original condition, and is fully functional. The W.T. May Company warranty sticker is still in place on the bottom of the machine, indicating its original 12 month parts and labor warranty.
Deitzgen's ESR-1 and Sperry/Remington's 1259S, Both OEM versions of the Casio FX-1
Along with being marketed world-wide by Casio, the same Casio-manufactured machine was marketed through the German firm Deitzgen, as the model ESR-1. The Deitzgen ESR-1 version was identical electronically and functionally, only cabinetry color, badging, and keyboard color changes differentiated it from the FX-1. Sperry/Remington also marketed a version of the FX-1 as the Remington 1259S, with only differences in color scheme (cabinet, keyboard bezel, and keycap colors). Casio later introduced an update of the FX-1 called the FX-2, with the sole difference being that the Nixie tube display of the FX-1 was replaced by vacuum-fluorescent tubes (reducing the cost). From a circuitry point of view, the FX-2 likely shared the same calculating engine. Only the display-related driver and power supply components were changed to support the lower current/voltage requirements of the VF tubes. At this time, it is unclear exactly when the FX-2 was announced. It is unclear if Casio's OEM customers (Deitzgen and Sperry Remington) offered their own branded versions of the FX-2.
The Casio FX-1 Opened Up
The FX-1 is built with three circuit boards which stack one atop another. The bottom-most board in the stack consists of all of the calculator logic. This board contains a whole slew of various small and medium-scale integration IC's made by NEC, Toshiba, and Hitachi, serving as glue to integrate the 5 LSI IC's made by NEC which make up the main calculating engine of the machine. The second board contains the tiny Nixie display tubes, an LSI decoder/driver IC for the Nixies, and various discrete components as support for driving the 12-digit display. The top-most board consists of power supply electronics. The keyboard assembly, based on magnetically activated switches, connects to the calculator board via an edge connector, and the power supply board and display board plug into edge connectors at the rear of the machine which have hand-wired connections to the calculating board.
Nixie Tube Display Detail
The FX-1 has quite a compliment of scientific functions, some of which are uncommon even on today's advanced scientific calculators. Among the functions the machine provices are sine, cosine, tangent, arctangent, and hyperbolic sine and cosine. The machine also has keys for conversions from degrees to radians and back, square root, cube root, conversion of degress/minutes/seconds to decimal degrees, natural logarithm, base 10 logarithm, ex, and ax functions. The machine has an accumulator-style memory register, with [M+] and [M-] keys, a [MR] memory recall key, and an [MC] memory clear key. It also has a separate store/recall memory register. The [K IN] key saves the number in the display into this register (not a constant, more on function of constant later), and the [K OUT] key recalls the content of the K register. The K register is retained through all but the power being turned off. The machine has a capacity of 12 digits. It can be set for full floating decimal point, or in fixed positions at 0, 2, 3, 4, 6 or 8 digits behind the decimal point, via slide switches at the left side of the keyboard panel. In fixed point mode, the machine can be set to round off results to the least significant digit displayed, or drop any digits behind the last displayed digit. The machine has an automatic constant feature which is activated for multiplication or division by simply pressing the function key twice. e.g., to repetitively multiply by two, starting with one, you would press " [X] [X]  [+=] [+=] [+=] ...". Also included in the compliment of keys is the [CHG Sign] key for toggling the sign of the number in the display, and an [EXC] key to exchange the display and operand register with each other. While the scientific functions of the machine are quite admirable for the day, there are a number of restrictions upon some of the functions which are interesting, indicating either a limited amount of ROM space to store the programs for the functions, or other restrictions in the internal logic of the machine. For example, the ex function will not allow an operand of any more than 9.99999999999, and the cube root function will not allow an operand of more than 999999.999999. Any larger operands will cause an error condition, even though the actual result is still within calculating range of the machine. The an function is also interesting, and unique. The first number (the number which is to be raised to a power) is entered, then the [an] key is pressed. Then, the keyboard accepts a SINGLE digit, and raises the first entered number to the power specified by that digit immediately. If any other than a key from zero through nine is pressed, the an operation is canceled. Basically, this means that this function can raise numbers from the 0th power to the 9th power only. While the operation is in progress, you can see that the machine simply does successive multiplications to calculate the result.
Error/overflow detection on the machine is OK, but not totally comprehensive. Taking the square root of a negative number yields a negative result with no error indicated. There is no dedicated error or overflow indicator, with the machine opting instead to clear the display to all zeros and locking the keyboard. Pressing the [AC] (all clear) key will clear the entire machine (except the K register) including any error lock condition. Pressing the [C] key clears any entry currently in the display.
Keyboard Layout of Casio FX-1
It is quite obvious when using the machine that it relies on ROM-based
programs (sequences of keypresses) for performing some of the scientific
functions. An example is the cube root function. The machine can take up
to 16 seconds to perform some cube root functions, while square root (likely
implemented directly rather than as a sequence of programmed
keypresses) takes a little over 1/2 second. During execution of the cube
root function, it is apparent that some kind of successive approximation
algorithm is used, as the displays cycle through patterns of digits which
converge on the result (see VIDEO). Some of the trigonometric functions can take
upwards of 6 seconds to complete. While some of the slower scientific
functions are being performed, the Nixie tube display digits dance in an
almost mesmerizing pattern. The standard four math functions operate as
is typical for machines of this vintage, with the ubiquitous
all nines divided by 1 benchmark taking about 1/2 second.