Ed Leofsky is, according to those who know him, a "meek, rather mild-mannered" process engineer at the Fairview, Pa., Electromagnetic Division of Spectrum Control Inc. But for 12 years, Leofsky lived with a thorn in his side.

Each week, Leofsky's division makes use of some 75,000 tubular ceramic capacitors, which are manufactured at Spectrum's Material Science Division in near-by Saegertown. The Electromagnetic Division solders terminals to the capacitors, which are then used in the electronic filters that are the company's main product. If, for some reason, the solder won't stick, Leofsky has a problem. Maybe thousands of them.

"It had been an intermittent problem, a serious problem, for as long as I'd been here," Leofsky says, "but we'd never gotten to the root causes." When soldering defects showed up, he had to have every item inspected with a microscope, a tedious, time-consuming process. Occasionally, Leofsky could fudge the manufacturing process enough to use poor-quality capacitors; but that, too, wasted time. Although he had discussed the situation over the years with H. Jack Baker, vice-president of engineering at Saegertown, the two had arrived at no solution. Finally, Leofsky had come to think of poor solderability as falling in the same category as death and taxes: not pleasant, but not up for debate.

Then, a year ago, two things happened. With a good deal of fanfare, Spectrum president and chairman Thomas L. Venable announced the advent of a new, companywide commitment to quality. Meanwhile, in Leofsky's department, the reject rate began to climb. By October, nearly a third of the capacitors wouldn't accept solder.

That was the point at which quiet Ed Leofsky decided he had had enough. Emboldened by the quality credos, classes, and posters that were cropping up all over Spectrum, he wrote a scathing memo to Baker and sent copies to everyone he could think of. "The albatross is upon us," he announced. "Rejects of . . . incoming material jumped from 3% in August 1984 to 32% in October. . . . Approximately 50% to 75% of engineering time is now being . . . diverted [by this problem]."

"At the time," Leofsky recalls, "I was thinking, 'Let's see if there's anything to this quality program."

"What he did," remembers Venable with a rueful smile, "was shut that department down -- and it was a department that accounted for 35% to 40% of the sales of the group."

Only rarely is the cost of poor quality -- and the cost of upgrading a company's quality standards -- thrown into such stark relief. But if Venable was surprised by Leofsky's dramatic action, he had to be pleased as well. It was just such an attitude that he and the other top managers at Spectrum had set out, some months before, to cultivate.

Spectrum, headquartered in Erie, Pa., was founded in 1968 by Venable, Glenn L. Warnshuis, and John R. Lane, three engineers who had met at Erie Technological Products Inc. In 16 years, the company grew from a $300,000 start-up housed in an old hardware store to a solid, $22-million public company. Today, Spectrum has four manufacturing plants and some 1,500 customers, including the likes of IBM Corp. and Hewlett Packard Co. For the past three years, it has reported after-tax returns of about 10% of sales.

In the early days, quality wasn't an issue. Venable and Warnshuis designed and built Spectrum's sophisticated filters, while Lane marketed them. "There wasn't any point in making them wrong," Venable says with a chuckle. But, as the company began to prosper and grow, that kind of hands-on responsibility fell by the wayside. "You probably lose control somewhere between 15 and 20 people," he reflects.

Like most manufacturers -- and like most businesses -- Spectrum began to operate on the philosophy of acceptable-quality levels, or AQLs. The company regularly checked a sample of the product, then shipped the whole batch, so long as the number of bad units fell within accepted limits. If there were too many bad ones, the lot was rejected, or subjected to 100% inspection, an expensive process.

Then, slowly, Spectrum's marketplace began to change. A Japanese company, Murata Manufacturing Co., purchased Erie Technological Products (now Murata Erie North America Inc.), with which Spectrum competed, and raised the specter of Japanese-style quality. Several of Spectrum's customers began to make noises about quality as well. "About two or three years ago," says Venable, "Hewlett Packard said that they were going to switch to the idea of 'zero defects' -- no defects in any inbound materials." Soon IBM was joining the chorus -- and implying, Venable remembers, that a business hoping to remain an IBM supplier better begin thinking seriously about quality.

Equally sobering were developments within Spectrum itself. "Our sales and marketing people began to tell us that we weren't very good at meeting schedules," says Venable, "and, in certain product lines, the reject rate began to climb."

Venable and other Spectrum managers began to assay likely strategies for attacking the newly discovered issue. They took a look at some Japanese quality techniques, but decided that the techniques would clash with the company's laid-back, American style. They bought 40 copies of Quality Is Free by management consultant Philip Crosby, a book that IBM had been pushing, and passed them out. They also bought and studied some videotapes featuring W. Edwards Deming, the dean emeritus of statistical control of quality. Surveying the options, Spectrum's managers came to the conclusion that the world was divided into Deming and Crosby camps, and that newcomers in search of quality were expected to enlist in one or the other.

Deming, a physicist-turned-statistician, was one of the principal inventors of statistical control, an approach that relies heavily on mathematical models to achieve and maintain quality levels. He was responsible for so many Japanese manufacturing breakthroughs that Japan's highest industrial award is known as the Deming Prize. Since the late 1970s, he has been accorded more attention and honor at home; one reviewer recently called him "the father of the Third Wave of the industrial revolution."

After seeing some of his tapes (which are produced by Massachusetts Institute of Technology), Spectrum ordered the entire package: 14 tapes, running some 16 hours, and costing $15,000. The company's top management endured the tapes at its monthly staff meetings, finding them provocative but ultimately disappointing. "It's simply Deming standing in front of a chalkboard, talking," says Venable. Spectrum was familiar with statistical control, had set up manufacturing systems compatible with it, and even practiced it "in a half-assed sort of way," but Deming's approach seemed somehow misdirected. "He says, 'Get your four statistics," says Venable. "Well, hey, we don't have any. And he's so profound that he becomes obscure; he'll say something like, 'Discuss third-order regression analysis. . . .' You just try telling that to some guy out on the shop floor. When we were through watching the tapes," Venable concludes, "we thought, 'Great philosophy . . . but what do we do?"

Crosby, once the director of quality for ITT Corp., seemed to warrant closer inspection. In July of last year, Venable and 16 of his top people boarded Spectrum's twin-engine Beechcraft Super King Air for Winter Park, Fla., the headquarters of Philip Crosby Associates Inc. There, for the next five days, they walked, talked, slept, and ate Crosby's approach to quality.

Crosby's pitch was "managing" quality: He was less interested in the details of execution than in giving management tools to create what he called "a quality environment . . . a new quality culture." His three books -- Quality Is Free, Quality Without Tears, and The Art of Getting Your Own Sweet Way -- are cracker-barrel collections of truisms, insights, and homey examples that boil down, in the minds of many, to a single, simple notion: Why not do things right the first time?

"He charges a helluva lot for that bit of wisdom," says Venable, pointing out that Spectrum's week of instruction cost $46,000. "We used to joke that he could just print that on three-by five cards and pass them out on the street."

Essentially, Crosby suggests that precise requirements be set for every business task, and that those standards be met each and every time. If problems occur, in either performance or product, permanent solutions must be found as soon as possible; temporary fixes won't do. And companies must track the "cost of quality" -- the price of conforming to requirements, as opposed to the cost of non-conformance -- both as an incentive and as a measure of success. Crosby's strict adherence to these standards in his own company -- there are elaborate housekeeping rules, and each instructor carries lists of "daily requirements" -- has earned him the nickname, among some students, of the Quality Fascist.

At Winter Park, Venable and his team found much to be skeptical about. They disliked the evangelical tone of Crosby's Quality College, the rote presentations, and what seemed to them like a lack of manufacturing expertise among the instructors. But the team also found much that was thought-provoking, and they attempted to adapt it to their needs. "We got a couple of coolers, and about 10 cases of Beck's and Heineken," Venable remembers. "Each evening, after class, we'd settle down with flip charts and beer, and figure out what we'd learned that day -- and how we could make use of it back in Erie."

Crosby's evangelical approach paid off in at least one major way: It destroyed the shibboleth of AQLs. "I think the principal benefit," Venable says, "was that it convinced us that, given a structure, it was possible to work toward zero defects, toward error-free performance."

Armed with such convictions, and fortified with $27,000 worth of Crosby's tapes and texts, the quality converts headed home. On August 16, at a gala kickoff party, Venable posited his and his company's new creed. "We are committed to quality performance," read the statement he had written for the company. "As an organization -- and as individuals -- we will continually seek out the specific needs of those who depend on us. We will then consistently satisfy these needs by doing everything right the first time."

The event was videotaped for the benefit of future employees, and Quality Education System (QES) courses were scheduled for every employee. Leaflets, posters, and "Quality Response Process" memos proliferated like mushrooms after a heavy rain; and Crosby's catchwords -- requirements, conformance, nonconformance, prevention, error-free -- entered the vernacular. Venable's plan was to use Crosby's razzmatazz and routines to get things moving, then rely increasingly on Deming's techniques to control the process -- modifying both, whenever it seemed necessary, with approaches of Spectrum's own design.

Some of the changes came easily, such as paying closer attention to customers' schedules. In the past, the company had often shipped its components too early, and the customers simply shipped them back. The cost of such errors, says Venable, was significant, particularly in the case of overseas deliveries -- "$150 to $200 for transshipping, and $300 for paperwork." At the other end of the pipeline, Spectrum installed new order-entry checking systems, "so we've seen a tremendous improvement in our error rate there."

For the most part, though, the improvements came slowly. "Easy?" snorts one worker in the Electromagnetic Division. "It was like giving up smoking and drinking, plus going on a diet -- all at the same time." Changing the habits and attitudes of Spectrum's workers was hard enough. But a thoroughgoing approach to quality involved the company's vendors and customers as well.

There was, for example, the matter of the bushings -- small, threaded items used to connect glass-sealed filters to other devices. The bushings were manufactured by three screw-machine suppliers, inspected by Spectrum, sent to a plating vendor, and, once plated, inspected again. At that late date in the process, some 50% were rejected.

"After you'd gone through the QES classes," says Electromagnetic Division unit manager David Weunski, "you were supposed to go back to your unit and think of things that had been giving you problems over the years. . . . This one, of course, leaped out at me."

The solution, however, did not. Only after endless hours of brainstorming and conferences with suppliers did Weunski hit on a strategy. During the initial inspection, he realized, Spectrum employed gauges that indicated only when the bushings exceeded the correct dimensions of the finished product; not until later, after another layer of metal had been added in plating, did other problems show up. So Weunski ordered $7,000 worth of new gauges, one set to measure the raw bushing and another to measure the plated one, and donated duplicate sets of gauges to his vendors. "Before," he says, "we would probably have put the burden of buying the gauges on them. Now, the attitude is much more cooperative." And the early results, he adds, are dramatic. "When all of the gauges are in place, we could be talking about a doubling of productivity."

Then there was the matter of Department Number Nine at the Electromagnetic Division, which produces, among other things, shielded windows. These windows -- artfully crafted panels of dark, curved glass that are fastened to the front of a computer screen -- absorb the six or seven watts of radiated energy produced by some computer terminals, and thus prevent anyone from "reading" the screen's information at a distance. But they are inordinately difficult to manufacture. Composed of layers of glass, wire screening, and laminating materials, they tend to delaminate when exposed to temperature extremes. "At worst, rejects were running as high as 15%," says unit manager Cy Ley.

Although Number Nine already had been wrestling with the issue, Spectrum's quality initiative pushed it to take some radical steps, such as changing vendors. For instance, Homalite Inc., of Wilmington, Del., had once provided the parts for the plastic laminate, but had lost the contract to another supplier; then, when quality became a top concern, it got it back. "Basically," says Homalite general manager Rod J. Field, "we lost them on price, but won them back with quality.They sent a four- or five-man team down to review what we were doing here -- and then we were back in business."

Department Number Nine supervisors have also become aggressively receptive to suggestions from line personnel. "No one is really an expert except the person who's out there building that window," Ley concedes. "One of their suggestions actually increased our productivity by something like 50%."

The net effect: a scant .08% reject rate on the newest line of windows. "Because of the dramatic improvement," observes Venable, "we were actually able to reduce our pricing on this product line."

Then, of course, there was meek Ed Leofsky and his shutdown of the soldering section. Virtually overnight, in the new culture, Leofsky became a minor celebrity, and substandard solderability became a major issue. The Material Science Division focused on the problem for a month, then sought help from the surface chemistry laboratory that is operated by Lord Corp., also of Erie. Lord found surface contaminants that could have been causing the grief, but left it to the division's engineering staff to locate the source of the contaminants and eliminate them.

"That," Baker notes, "is when it really became frustrating." No sooner had one source been identified and dealt with, than another, and then another, would appear. The materials division seemed to be battling ghosts.

"It took two months," explains Leofsky, "before they began to make progress in pinpointing the troublesome areas. Since then, they've been tuning and fine-tuning all of their procedures." It was, he adds, an expenditure of manpower, capital, and concern that, in the past, would have been unthinkable. "We would never have gotten beyond justifying the return on investment," he says.

Overall, there are few people, processes, or products that haven't in some way been affected by Spectrum's quality crusade. There is now a vendor-selection committee, for example, and the number of active vendors has been trimmed by 8%. The company is also more demanding of customers. When it felt that one client's specifications for a filter used in the B-1 bomber were unattainable, it said so, and lost the work, but promptly got it back when the competitor that got the job discovered (and proved) that the unit couldn't be built as designed. Not even the company's outside directors have escaped the reeducation process: Venable recently asked several of them to attend Crosby's Quality College.

Tom Venable, for his part, is happy with the results, despite the difficulties. "In our first quarter of Quality Response Process operation," says Venable, "we've seen a 75% reduction in sales returns and allowances; if you annualize that, you're looking at savings of something like $767,000." Even more telling is Spectrum's profit-sharing balance. Believing that employees should have a fiscal, as well as a psychological, incentive to get involved in the program, Venable earmarked about half of the savings realized for the company's profit-sharing plan. Last year, management had put $150,000 into the program, but, high on quality, had budgeted $525,000 this year. Now, observes Venable," We have the feeling that it's going to be quite a bit higher -- more like $1 million-plus."

Best of all, maybe, Ed Leofskymay finally be able to stop worring about solderability. On March 12, 1985, 13 out of 14 lots of the tubular capacitors passed incoming inspection, and that level of quality has held steady ever since. On March 18, Leofsky sat down and wrote another memo: "We offer you and your team congratulations for the tremendous progress that has been made."

"Life here is a lot more pleasant these days," he adds with a smile.