Reference

Sidebar 2: Integrated Circuits

In 1950, the US Navy's Bureau of Aeronautics needed to build complex, reliable, serviceable weapons systems, but were hampered in this by the tube technology of the time. They thus set up a study of automated tube circuit assembly. The resulting prefabricated circuits were fine, but by the time they were ready, the crucial development of making transistors reliably and cheaply by making them "epitaxially"—ie, by growing doped layers on crystals rather than by hit-or-miss mechanical assembly—made several million dollars' worth of fine tube-based engineering obsolete.

By 1957, it was obvious that while transistor cases were already small, tens of transistors could be fitted in the same space usually occupied by one. In 1958, Jack Kilby of Texas Instruments went further by discovering how to "grow" a useful part of a whole circuit—not just interconnected transistors, but resistors too—on a piece of silicon and fit it in a tiny case. Today, the same Texas Instruments—introducer of Kilby's "integrated circuit"—remains one of the biggest players in the cutthroat world of IC manufacture.

Since about 1970, silicon-based ICs have been the increasingly-taken-for-granted alternative to laboriously wiring up individual ("discrete") parts to perform common functions. They permit more electronics in a given space at lower cost, and, ideally, use fewer material resources. A modern IC costing $2 to make would cost many times the price—at least $25, and as much as $250 if audiophile-grade—if made from discrete parts, even using modern automated, surface-mount assembly and test. Also, equipment using ICs can be built with less skilled labor. Because all the transistors in an IC share the same substrate and doped layers, the performance of ICs can also be more consistent, and IC-based designs can usually be built or repaired with only simple testing and basic knowledge. Altogether, these advantages make ICs an attractive approach for makers.—Ben Duncan