Early Electronic Calculator Casio AL-1000 Calculator Technical Description: Architecture

1. Elements and Structure

The AL-1000 is a digit-serial decimal number processor controlled by a hardwired state machine and driving a multiplexed display. The digits of decimal numbers are processed in serial, with each 4-bit BCD digit being processed in parallel. Primary data paths are 4-bits wide. The control state machine is quite well defined, which assisted greatly in making it amenable to reverse engineering.

Figure 1 is a block diagram of the AL-1000. Following this are brief descriptions of the various elements.

The names for flip-flop elements are taken from the labeling on the printed circuit boards. Some of the names are non-alphabetic symbols on the PCBs and have been renamed, such as "÷" with "FDIV". Most of the names are a readily apparent abbreviation of their purpose. A few, such as the "T" flip-flops have remained ciphers as to the source or meaning of their name, although their purpose has been decoded. Names for unlabeled elements and signals have been created as seemed appropriate.

Figure 1: Casio AL-1000 Calculator Architecture

2. Memory Organisation

From a user perspective, the calculator provides two operand registers, 4 memory registers and 30 locations of program storage. These are all stored in a small core memory.

The two operand registers (X and Y) and two of the memory registers (M1 and M2) are each 14 numeric digits, a sign digit and a digit to hold the decimal point position. The other two memory registers (M3 and M4) are 7 numeric digits and a decimal point position each (no sign). The contents of the X register appear in the Nixie tube display, the Y Register is the '2nd operand'.

The program storage is comprised of the equivalent of another two full registers. The numeric and sign digits are the instruction storage locations. The decimal point position digit is used as the program counter during program execution. The PB flag is used to select between the two sets of instruction storage. The program storage can be utilised either as two programs of up to 15 instructions each or as 1 program of up to 30 instructions.

The functional layout of memory is shown in table 2a. It might seem sensible to implement this in core as 4 bit-arrays of 16 by 8 or 16 by 7. However, the designers chose another layout: 4 arrays of 14 by 8, as shown in table 2b. Each 16-digit register occupies two 8-digit columns. This probably simplified the address decoding logic slightly as there are no wasted cores, nor is special logic needed for the absent upper digits of M3 and M4. The ordering of registers in columns is muddled, perhaps for the sake of more direct PCB traces, as it doesn't matter as far as operation goes.

All digits are 4-bits, numeric digits of numbers are encoded in BCD (1248) format, the decimal point position and program counter digits are base-16. Program instructions are also base-16, using the same coding as that which the function keys are encoded into for storage in the U latch, as presented in table 2c. Programs are entered by using the numeric keys to enter the appropriate code for a function. For code values greater than 9 the decimal point is pressed to turn on the '8' bit. Code 13 (or '5.') is used internally during the square root procedure and is not intended for use by the user.

3. Printed Circuit Board Contents

The various elements are distributed amongst the printed circuit boards as follows: