June 1969 Electronics World
Table of Contents
Wax nostalgic about and learn from the history of early electronics. See articles
Electronics World, published May 1959
- December 1971. All copyrights hereby acknowledged.
Mr. Lothar Stern was a fairly prolific publisher of books and magazine articles during his long career at Motorola. What remains of the Motorola brand these days is mostly associated with cellphones, with the venerable Razr model being the most famous. China-based Lenovo now owns even that aspect of Motorola (after a short-lived ownership by Google). In its heyday prior to the 2000s, Motorola was a major manufacturer of analog, RF, and digital semiconductors (including microprocessors), radios, televisions, telephones, vacuum tubes, medical equipment, and more. It is interesting to note that the issues referenced in this article regarding technical terms are note even relevant anymore.
See Lothar Stern's article titled "Silicon Controlled Rectifiers - New Applications in the Home," in the October 1963 issue of Electronics World.
Some "Technical Terms" Aren't
Engineer, teacher, editor, and author, Lothar Stern has been involved in communicating electronic technology for the past 18 years. After receiving his B.S. in television engineering, he worked as an electronics engineer and instructor of electronic engineering and servicing for several years. In 1961 he joined the Semiconductor Products Division of Motorola as Manager of the Technical Information Center, charged with the responsibility of conveying the changing aspects of semiconductor technology to semiconductor users. He is the author of several books.
By Lothar Stern, Manager Technical Information Center, Motorola Semiconductor Products Inc.
It is deplorable that a field as technically precise as semiconductor electronics should be so permeated by ambiguities in vocabulary that even those directly involved are finding it increasingly difficult to communicate. In part, this is an unfortunate but perfectly understandable condition in a rapidly advancing technology. For example, we can no longer expect the word "transistor" to convey a definite meaning without modifying it by the adjective "bipolar" or "field-effect." Nor can we speak meaningfully of an "integrated circuit" without qualifying it in terms of structure (i.e., monolithic, hybrid, etc.). But, in addition to the word explosion brought on by the expansion of functions and processes, we find ourselves victimized by the proliferation of terms without clear-cut meaning or value. Such a term is the overworked acronym "LSI."
Anyone connected with electronics in any way probably knows that LSI stands for "large-scale integration." It's a good bet, however, that out of any group of three engineers chosen at random, no two will define it the same way.
There's no record of how this term got its start. But so ambiguous has the word become that one executive in the industry has disgustedly defined LSI as "large-scale insanity," while an engineer has been quoted as saying that, "LSI is anything more complex than anything we can make today."
There have been attempts, of course, to define LSI more quantitatively. One group proposed that any monolithic circuit composed of more than "100 gates" should be called LSI. This was amended to "100 equivalent gate functions" (whatever that is), when it was pointed out that linear circuits, too, might become very complex.
Finally, when challenged with the question of what a monolithic circuit of only 99 gates might be called - well, what is more logical than to coin another completely ambiguous term, Medium-Scale Integration (MSI). Thus, MSI might be defined as any circuit larger than small-scale integration, but smaller than large-scale integration.
The big problem of placing numbers on LSI is, of course, the changing technology. While a 100-gate circuit might have seemed difficult three years ago, it certainly is within the realm of practicality today. And within a year or two, such circuits are likely to be very simple in comparison with the state of the art.
Another school of thought has it that, to achieve LSI-status, a circuit must have two or more layers of interconnecting metallization. The rationale for this line of reasoning is that no circuit with the interconnecting metal on the same plane as the components themselves can achieve a high enough order of complexity for LSI. The fallacy here is that multilayer metallization technology is rapidly becoming so routine that it is being adopted, for economic reasons, even for some of the not-so-complex devices. Obviously, this definition won't stand up either.
One might be tempted to avoid the entire LSI/MSI hassle by ignoring the terms altogether, were it not for the fact that provocative headlines and ad copy continually remind us that something unusual and very exciting is happening in our technology. And how the reader interprets this liberally used term is usually quite incidental.
In the absence of any industry agreement through the EIA, we proffer the following definition, which is currently endorsed by and practiced at Motorola Semiconductor.
LSI is the simultaneous realization of large area circuit chips with optimum component packing density, for the express purpose of reducing circuit costs by maximizing the number of system interconnections performed at the chip level.
Note here the complete freedom from the "tyranny of numbers." Not a hint as to process or manufacturing technique. Simply a statement of goals that, to its credit and its detriment, can be construed to fit any complex integrated circuit - including those envisioned for tomorrow as well as those in existence today.
Posted May 9, 2017