The more things change...
Twenty years ago, the ATE industry was cyclical, interference was troublesome, standards were emerging, design-for-test was becoming a necessity, and editors were looking back and ahead.
Staff -- Test & Measurement World, 9/15/2001
 |
|
1981 |
With an issue dated “Fall 1981,” Test & Measurement World first arrived on engineers’ desks. Dedicated to “the engineer or manager responsible for selecting and implementing test and measurement equipment for the electronics OEM and its end-user industries,” the magazine promised to “guide you through the latest advances in test and measurement technology.” Today, 20 years later, our mission remains much the same.
Upon reaching this 20th anniversary, we decided to look back to see how issues in the test industry have changed—or not changed—through the years. Here, we present some of the highlights from our first year of publication, along with our own comments. |
 |
|
1986 |
Fall 1981:
ATE: Still the Promised Land?
Caught in the throes of the worldwide recession that has cooled the semiconductor industry, ATE firms have reported flat sales and reduced profits for the first three quarters of 1981. But many companies still feel that the slowdown won’t impact the long-term health of the ATE marketplace.
Sound familiar?
Then and then: 20 years ago, Test & Measurement World editors tried their hands as historians, looking for the first test engineer:
Fall 1981:
Then and Now: A Chronicle of Ancient Test Instruments
The science of electricity began to dominate the development of instrumentation some 150 years ago. For many years, electricity remained a novelty item, but by 1784 George Adams, instrument maker to King George III, was sufficiently impressed by the phenomenon to aver that “the science of electricity is now generally acknowledged to be useful and important . . . [A]t a future period, it will be looked up to as the source from whence the principles of natural philosophy must be derived.”
Test and measurement sprang up as a natural adjunct to the discovery of electricity, and one of the earliest instruments used to measure the quantity of electricity was the balance discharge electrometer.
 |
|
1991 |
Fall 1981:
In-Circuit Tester Identifies Faults Automatically
. . . ATE suppliers are becoming more aware of the entire PCB manufacturing process, especially the need to upgrade production as well as test throughput. Hopefully, this recognition is the start of a trend toward better integration of manufacturing and testing resources.
It’s an ongoing struggle. Today, test equipment vendors are still striving to use test results to fine-tune manufacturing processes.
Have you tried to have DSL installed lately?
January 1982:
Telephone Line Testing: Automated Approaches for the New Digital Technology
. . . If the telephone industry converts to all-digital switching as planned, it’s possible that almost all the subscriber lines in the United States will be tested automatically by the year 2000. And if this happens, the customer service call may become as much of an anachronism as the hand-cranked telephone.
Logic analyzers were going strong nine years after their introduction, and they still are, although their developer has a new name.
January 1982:
Logic Analyzers in 1981: Processing and Stimulus for MPU Development, Portability for the Field
Although the logic analyzer is less than a decade old (Hewlett-Packard brought out one of the first in 1973), it has already become a mainstay of the digital test and
design bench.
 |
|
1997 |
The crotchets are gone, and the standard lives on. And is Rubik’s cube really that old?
January 1982:
HP-IL: A New Interface for Low-Cost, Portable Measurement Systems
IEEE Std. 488-1978, also known as the general purpose instrument bus (GPIB), has been fine tuned and updated since it first became standardized in 1975, but it still has some crotchets. The document itself remains as tough to figure out as a Rubik’s cube . . .
Nonetheless, the test and measurement industry seems pleased with IEEE 488. Over 100 vendors now supply GPIB-compatible products, and countless users have hopped on the bus in its six years of existence.
February-March 1982:
The Evolution of Board Testing: In-Circuit Merges with Functional
Manufacturers designing a board test facility often deal with companies that supply both in-circuit and functional testers. With the trend toward combining the two approaches in one system, recommended test strategies may vary with the types of systems offered.
. . . Once easily recognizable, the distinctions between in-circuit and functional approaches are becoming hazy, primarily due to the growing sophistication of in-circuit test techniques. Furthermore, each ATE supplier means something different by the term “combined in-circuit and functional test system.”
Still a winning combination, and still not easily recognizable.
February-March 1982:
Where are we Going in Test and Measurement? How Will We Get There?
The evolution of test instrumentation technology can be traced by looking at front-panel controls. Taking a “trip down memory lane,” . . . Norbert Laengrich of Racal-Dana returns to the not-so-distant past when instruments still had dials, lamps and knobs—and then examines the mP-based instrumentation of the present, with its keyboards, CRTs and pushbuttons.
You’ll still find the occasional pushbutton.
Still shocking after all these years:
February-March 1982:
EMI Testing: New FCC Standards for Digital Equipment
The Digital Age might just as accurately be christened the Age of Interference, for digital electronic products transmit high-frequency signals in much the same way that radio waves are transmitted.
 |
|
1999 |
April 1982:
Eliminating RFI at the Design Stage
1. Keep fast rise-time digital signals at as low an impedance level as possible, with circuits running over a ground plane sheet on a PCB, where feasible. That is, treat the leads carrying these signals as transmission lines so they don’t act like antennas.
2. Apply extra line filtering to motors.
3. Use some amount of current limiting in power supplies; brute-force filters have high charging current spikes. By sacrificing a little energy, input resistance or inductance on the power supply filter can limit peaks.
4. Employ double-shielded power transformers (shielded primary and secondary) in the design, especially in the presence of switching-mode regulators.
5. Place power-line RFI filters where the power cord enters the chassis.
6. Use shielded power cords, with the shield tied to the powerline ground (green) and to the cabinet.
7. Use shielded data cables.
Still good advice, but I’m not sure we can afford the energy sacrifice mentioned in step 3.
Test & Measurement World looks for the first ATE system.
May 1982:
Then and Now: Computers Meet Test Equipment and ATE Is Born
What was the first automatic test system? It’s hard to pinpoint the exact birthdate
of ATE, which was the product of many creative minds. But Don Glancy, West Coast editor of Test & Measurement World , feels the honor of being the first automatic tester to incorporate a computer should go to the Nortronics DATICO Serial Processor (SP)-5. This machine was originally conceived in 1959; a laboratory model was constructed in 1960, and the first production model was delivered in 1963.
 |
|
2001 |
September 1982:
The UNIX Operating System: To Be—or Not to Be?
With ATE manufacturers constantly searching for ways to respond to competitive product and pricing pressures, one sure way to gain an advantage in the marketplace is to offer software products that are compatible, transportable and long-lived. However, within the minis, micros and mainframes necessary for ATE, the diversity of assembly languages and the incompatibility of operating systems . . . have kept the vendors far from their goal. Thus, an application package such as GenRad’s CAPS digital logic simulator functions only on its original host computer, the DEC PDP-8, and only under the O/S 8 operating system.
. . . Excellent ATE software packages, such as Instrumentation Engineering’s
MICROSYM, Computer Automation’s SPRINT and Teradyne’s VISIONS, may some day be obsoleted by their inability to adapt to new hardware.
. . . The one attempt at a universal test language, ATLAS, has had only limited success in cleaning up the endless proliferation of ATE software.
. . . [T]he latest attempt at a fully portable operating system shows great promise in its ability to make software transparent to CPUs. Although Bell Laboratories’ UNIX is not the answer to all our software problems, it does provide a convenient, portable operating system on a growing number of mainframes, minis and micros.
. . . Possibly, some of the hardware suppliers who benefit from their product’s uniqueness may oppose the spread of UNIX . . .
Still looking for convenience and portability—Linux, anyone?
October 1982:
The Present and Future of VLSI and ATE
Paced by the growth of CAD and the advance of IC fabrication technology, VLSI circuits are growing denser and more complex. Testing those devices is difficult, since the number of I/O pins is minimal, and internal circuit visibility is poor. In the future, VLSI designs will have to incorporate a large measure of self-test capabilities and test equipment must adapt to changing technologies.
A statement that’s still true today: “In the future . . . designs will have to incorporate a large measure of self-test capabilities.”
Fall 1981:
Test Equipment Behind the Iron Curtain: Exports and Espionage
The test and measurement industry is not normally a political animal. Like most other high-technology business sectors, it is caught up in the never-ending cycle of developing, marketing and selling products . . .
Since the Soviet takeover of Afghanistan, however, a number of American business segments have collided with the amorphous but powerful interests of “national security.” Growing concern over Soviet military intentions has induced the U.S. government to impose stricter controls on trade with the Eastern bloc, particularly on exports of high-technology goods . . .
In Los Angeles, the U.S. District Court recently convicted two people of high-tech espionage involving substantial amounts of test equipment . . .
We still have spies, but the Iron Curtain is thankfully gone. T&MW
Rick Nelson received a BSEE degree from Penn State University. He has six years experience designing electronic industrial-control systems. He has served as the managing editor of EDN, and he became a senior technical editor at T&MW in 1998. E-mail: rnelson@tmworld.com.
| The men who started it all
The events of the last five years also include the deaths of David Packard (March 1996) and William R. Hewlett (January 2001), the founders of Hewlett-Packard. Without a doubt, the electronics industry owes these two men a debt of gratitude, not only for launching the test-and-measurement industry, but also for their steadfast belief in the value of people. |
THROUGH THE YEARS
T&MW took a long look at inspection
When Test & Measurement World got its start in late 1981, dimensional measurements seemed to garner the most attention. Readers could read articles about optical comparators, dimensional-measuring equipment, and how to make submicron linewidth measurements. Machine vision (MV) was covered only slightly as the magazine explained how to use robotics and MV to locate visible shorts in thick-film circuits. One author explained, “Micro- or minicomputer systems typically cannot process image data quickly enough to give real-time feedback on a variety of features.”
Five years later, in 1986, a T&MW article covered a new inspection technique that used a scanning laser acoustic microscope (SLAM). A SLAM could locate voids in the bonding materials between a semiconductor die and its substrate. Acoustic microscopy technique continues to find use in applications such as failure analysis.
T&MW also covered surface analysis and explained the operation of scanning-electron microscopes (SEMs) and transmission-electron microscopes (TEMs) for readers who might need to analyze the surface of a hard-disk platter, for example. Most electron microscopes provided images users could view, but two authors described how to transfer image information into a small computer. Modern SEMs and TEMs come with built-in computers that control the instruments and process their images.
Engineers started to become interested in 3-D measurements, and an article in 1986 explained how researchers had developed a 3-D camera. The camera technique never caught on, but today you can buy systems that use x-rays or lasers to quickly measure volumes of solder paste.
After 10 years, the inspection scene had changed to include scanning-tunneling microscopes—also called scanning-probe microscopes—as tools used to inspect submicron features on ICs. T&MW’s readers got exposed to more information about x-ray inspection techniques that could examine devices that had leads on 20- and 25-mil centers. But the speed and cost of x-ray systems put them beyond the reach of many potential customers.
By 1996, inexpensive PCs and easy-to-use software packages, combined with a cornucopia of small cameras, made MV systems easier to develop and integrate into a production line. The prices of x-ray systems had dropped, so more production facilities could use them to inspect for hidden defects. Still, x-ray systems took time to perform an inspection, so few systems found use on high-speed production lines. Test engineers also started to take advantage of lower-cost infrared inspection systems to locate defects and potential hot spots. In the mid ‘90s, more PC-based MV systems supplied connections that linked them with process and test equipment, and with corporate databases.—Jon Titus |
THROUGH THE YEARS
In 20 years, measurements haven’t changed
Looking through past issues of Test & Measurement World, I’m struck by the notion that while test engineers can take more accurate measurements faster then ever before, the basic measurements haven’t changed. If you think about the measurements you made 5, 10, 15, or 20 years ago, you realize that you’re making the same measurements. You still measure volts, amps, watts, seconds, and Hertz. In telecommunications, you still make the same measurements such as signal power, rise time, and bit-error rate. The same parameters that limited measurements 20 years ago—speed, accuracy, noise, time, and money—still limit measurements today.
In 1982, oscilloscope manufacturers pushed scopes with 60-MHz bandwidths. Today, mid-range scopes have 10 times that bandwidth. Scopes may run faster and can process data in ways that were difficult or impossible until perhaps 10 years ago, but they still make voltage measurements. And you still have to trigger a scope to get a measurement. You just have more trigger options now.
Engineers also use DMMs in the same ways they did years ago. In April 1996, we reviewed four 61/2-digit DMMs. The two our reviewer liked best are still in production. What’s interesting, though, is these meters defy the general trend of faster, better, cheaper electronic products. They still sell for the original price of $995.
Computers have brought productivity gains to test engineers. Computers are faster and easier to use then they were just last week, and PC-based instruments are easier to program than ever. Still, I hear the same complaints from engineers we heard years ago. In 1992, for example, I quoted an engineer who wanted a common set of programming commands for instrument cards from different manufacturers. Unfortunately, he’s still waiting.
In other ways, test engineers still use the tools they used years ago. Despite Ethernet, USB, IEEE 1394, and other computer buses, none has overthrown the IEEE 488 port or the serial port as king of instrument communications.
You also may think regulatory requirements have changed EMC measurements. Europe’s 1996 EMC requirements for the CE marking may have added many new measurements to electronics equipment, but those measurements still involve volts, amps, watts, and Hertz.—Martin Rowe |
THROUGH THE YEARS
Design and test struggle continues as RF becomes pervasive
Many things never change—for example, the need to choose the right tool at the right price for the job. In RF test, we reported in January 1986 that “Network analyzers serve to characterize . . . RF and microwave equipment . . . [V]ector analyzers offer some important advantages, but they cost roughly four times as much as scalar analyzers, which are entirely satisfactory for many purposes.” That’s still true today.
Some things do change, however. ATE performs RF tests once relegated to stand-alone instruments. We noted this trend in March 1996; the article “RF Testing with In-Circuit ATE” described the challenges: “. . . accurate RF observation is not possible because of the abnormal electromagnetic environment; just placing probes on or near the electronics changes enough of the circuit characteristics that any true representation of normal operation is impossible.” In September, we emphasized the importance of meeting the challenges: “You will need to understand how to apply new measurement techniques to high-volume production testing of complex digital cellular and emerging PCS systems.”
In the ATE arena, throughout our 20-year history we’ve reported on constant predictions of the obsolescence of various test approaches: “Both in-circuit testers and MDAs carry the seeds of their own demise—every board requires a bed-of-nails fixture,” we reported in February 1986, noting that fixtures are expensive. Consequently, “After years of being eclipsed by in-circuit testers, functional board testers are catching on again,” we added in April 1986. Today, in-circuit and functional testers still coexist.
Design-for-test has long been important. “As design cycles shorten and IC complexity increases, the interaction between design engineers and test engineers must grow,” we noted in June 1986, after reporting in March that “Although information necessary to generate comprehensive test programs for CAD-designed ICs and PCBs resides in their CAD databases, no universal technique exists to transfer that information to test-program generators.” By August 1991 there hadn’t been much improvement: “The links between ATE and EDA are confused in a flurry of official, de facto, and would-be standards . . .” Vendors of ATE and CAE tools are still debating the relative benefits of standards vs. the performance edge available from proprietary tools.
One approach that has fallen from favor during our history is incoming test. In January 1986, we reported that “. . . escalating electronic component complexity increases . . . the likelihood that any particular chip will contain a defect.” To extirpate those defects, we suggested customers perform incoming test, particularly for suppliers without a track record. It’s a tribute to suppliers’ success that incoming test is largely history.—Rick Nelson |
