Jerry D. Merryman, who was selected as the project manager, came up with the logic designs for the calculator in three days. James H. Van Tassel, who had been working on research and development of semiconductor components at Texas Instruments, also joined the team. Together, Kilby, Merryman, and Van Tassel began work on the calculator in October 1965.

On September 29, 1967, the three filed for a U.S. patent for the world's first handheld calculator. Although the machine did no go into actual production for three years, the prototype had been made. The calculator's use of integrated circuits was a departure from from anything that had been done before. "We called it large-scale integration, even though the semiconductor array incorporated only thousands of discrete devices," said Merryman. Modern arrays incorporate millions of devices.

The packaging of the integrated circuit posed a problem. "Most circuits had 14 or 16 leads. Ours had 120. We were on new ground. We had to find the chip's coefficient of expansion and some way of making reliable contacts, " said Van Tassel.

The quality of the two-inch wafers available at the time was not assured. "The chips had to be probed. When you found the good ones, you could hook them up. But then you had to sort through those and find the ones with enough good area in common," said Van Tassel.

Merryman's logic designs were brilliant, according to Van Tassel. The number of possible logic functions was limited by the power supply, which Kilby specified must use batteries. In addition, the display was a bit of a problem, said Kilby. The engineers finally chose a thermal printer, which used relatively low power, as the output device. The paper, which was made by NCR Corp. (Dayton, Ohio), had microencapsulated ink. "It was like carbon-less carbon. The paper was pressed against the back and the number was heated. Then you had to advance the paper," said Van Tassel.

The machine could add, subtract, multiply and divide. There was some automatic decimal placement, and input numbers could be as large as 12 digits.

The machine's dimensions were about 4¼ by 6¼ by 1¾ inches and it weighed about 45 ounces. However, the engineers had constructed the case from a solid piece of aluminum. "What you really had was a block of almost solid metal," said Merryman. The batteries were considerably heavier than current versions. Initially, the calculator used silver-zinc batteries, because the nickel-cadmium models available in 1967 were not very reliable. However, within a year of the calculator's invention, nickel-cadmium batteries had been improved enough so that they could be used in the machine. The batteries lasted for about four hours before needing recharging.

The model was not mass-produced immediately. Texas Instruments approached Canon Inc. (Tokyo) and arranged to coproduce a pocket calculator completely built with Texas Instruments parts. In April 1970, the Pocketronic appeared on the Japanese market; it was a four-function, entirely electronic calculator that retailed for about $400. The machine was marketed in the United States in the fall of that year.

Kilby, Merryman, and Van Tassel were presented the ASME Holley Medal in 1989 "for a great and unique act of an engineering nature of timely public benefit." Kilby, who produced the first integrated circuit in 1958, now works as an independent consultant. Merryman is currently a Texas Instruments Fellow. Van Tassel is vice president of the microelectronics division of NCR Corp.

None of the calculator's three inventors said they could foresee in 1967 the status the calculator would hold today. "The calculator project was intended to show what we could do with integrated circuits and to broaden the market for them," said Kilby. "and we did go through and build the first calculator."
 

Irene Kim, "Functions at Your Fingerprints." Mechanical Engineering Magazine. Vol. 112, No. 1. January 1990.
Published with permission of the author.

French Translation made by Anna Chekovsky