By Neven Karlovac
Integrated System Design
Posted 02/26/01, 06:10:11 PM EDT

It's a frustrating paradox hardware engineers constantly face. You're tasked with designing an innovative product using the latest in component technology. And, although the discovery phase is time-consuming and complicated — requiring you to investigate the latest chips and algorithms, and research prior art and patent issues — you're also asked to deliver your design quickly in your company's hurry to "bring it to market soon, make it profitable fast and . . ." yada, yada, yada.

Under these conditions, you must gather as much information as you can in as short a time as possible. Problem is, there is no comprehensive repository of the type of information engineers need in the discovery phase — component information and specifications, reference designs, compatibility issues, proprietary issues, etc. Engineers, therefore, are forced to rely on visits from distributors and sales representatives, trips to libraries and calls to professional associations for the information they need. And, as engineers know, obtaining information this way can drain weeks of precious discovery time. Here's where the Internet should come into play, which brings us to the second problem.

Engineers should be able to conduct searches for the discovery-phase information they need on the Net quickly and easily. After all, search engines work perfectly well for the average Net user. If he wants an online travel agency, the typical Net-savvy individual simply types in "online travel agency," and he's off and running. And herein lies the problem: Most of today's search engines are keyword-based, and engineers require much more sophisticated search capabilities — the ability to find relationships between objects, search for specific attributes, drill-up/drill-down, etc. An engineer might need a frequency synthesizer for a receiver in a given frequency band. (Try typing that word string into a search engine; you might actually hear your computer laughing.) Engineers need to be able to conduct iterative searches for the information they need, searches in which they can specify a series of complex parameters and mathematical relationships-"if/then," "at least," "must have," "maximum," etc. — and alter those search parameters on the fly.

None of the standard search engines meet the requirements of engineers' searches. Most are unstructured and keyword-based. Thus, an engineer searching for microprocessors who enters the keyword "microprocessor" into a typical search engine will receive hundreds of thousands of results, most of which will be completely useless to the engineer.

Search engines with Boolean operators offer a slightly more effective alternative for engineers. Boolean operators are search elements that allow the user to input relational, comparison or conditional expressions in a search command — such as "true/false," "greater than/less than," and "and," "or" and "not"— for more accurate results. An engineer could, for example, enter the command "chip not potato" into a Boolean-equipped search engine and filter out thousands of inappropriate results. Similarly, an engineer looking for a utopia microprocessor could enter "microprocessor and utopia" into an engine with Boolean operators for more accurate results.

However, even these more sophisticated engines, which offer advanced keyword searches with Boolean operators, fail for several reasons: First, they still return far too many or too few results. Even a Boolean search for chips that rules out potatoes may well return hundreds of thousands of Web pages completely useless to the engineer. Second, Boolean operators fail to enable engineers to conduct parametric- and rule-based searches for voltage requirements, compatibility and packaging. This information is built into a chipmaker's PDF datasheets, but standard search engines will not search these PDF documents. Even in the case of sites that have searchable databases, the standard search engines cannot extract the data from those databases and return them in keyword searches. An engineer who conducts a search for a chip with a given frequency that is "greater than x" will not likely find any direct matches, because the engine will not be able to search the datasheet that contains that information. Thus, even with Boolean operators, the engineer's task of scouring the Web for the best possible components is only slightly improved.

If forced to use a standard search engine to conduct early-phase component research for a new product, an engineer's best bet is to find an engine that allows Boolean expressions, and input as many of the desired elements (e.g., "microprocessor and DSP and...") as the search will support. Of course, this is only a first step in narrowing the vast amount of information on the Web and will not likely result in an exact match.

With product lifecycles shortening and manufacturers in an ever-increasing hurry to bring new products to market, engineers need the proper tools — intelligent search technology tailored specifically to their unique needs, one-stop access to the vast sea of information necessary to absorb during the discovery phase, and the ability to create diagrams and test component compatibility-to navigate through the complex discovery process and design a state-of-the-art, competitive product.

The engineering community needs tools, accessible from the Web, that will speed up and improve design-discovery time — and we need them soon.