- SPECIAL REPORTS
Beyond the component level, electronics manufacturing consists mostly of soldering: making, evaluating and testing solder connections. Over time, electronics manufacturing companies have increasingly relied on IPC-A-610 and IPC-J-STD-001 to define what soldering conditions are acceptable. And, as more and more former manufacturers turn to contract manufacturers, the customer base also has accepted these IPC documents as gospel. Too often, however, both the manufacturers and their OEM customers are deceiving themselves into accepting compromised output produced inefficiently.
IPC-A-610 and IPC-J-STD-001 continue to play a large role in the demise of American electronics manufacturing. Far from helping users become more efficient producers of high-reliability products, these two directives have fostered a culture of touchup and rework that inflates costs and product failures. This must change.
The fundamental failing of both IPC-A-610 and IPC-J-STD-001 stems from their origins. Basically, they grew out of defense contracting practices of the 1960s, especially the original “Zero Defects” practices born at The Martin Co. in 1961. Zero Defects (ZD) was an initiative to see if Martin could ship its missiles completely free of defects. Despite its name, ZD did not actually mean doing the job right the first time. Instead, each assembly was inspected to detect nonconformances that were then reworked. The reworked units were also inspected and any additional defects were reworked. And so on.
Reliance on inspection and rework is a highly inefficient operating methodology. Defense contracting, however, rewarded inefficiency. (In many cases, it still does.) The key is “cost-plus” contracts under which suppliers’ profits are a percentage of product cost. In the nondefense world, pay for inspectors and rework personnel comes straight off the bottom line. With a “cost-plus” contract, however, the customer reimburses the supplier not just for the costs of inspection and rework, but also a percentage of those costs. Finding and fixing defects can put a civilian company out of business, but Martin made more money by making more defects. More inspection and rework led directly to higher profits.
I am not suggesting that Martin deliberately created defects. But there was never any economic reason for pursuing efficiency and doing the job right the first time. An executive of one defense contractor in the 1980s actually told me that the company was not interested in activities that would have increased efficiency and reduced defects because that would not be good for profits. The defense industry has been reworking defects for so long that few employees believe that defects can-or should-be prevented.
The ZD mentality continues as standard operating procedure at most of the large defense and aerospace manufacturers. Time and again, I have been in plants where they might as well turn off the soldering machines because their operators rework every solder connection anyway. Often, the same connections are reworked more than once and, not infrequently, many times. Solder connections are inspected under microscopes and reworked until they resemble tiny jewels. Connections that are known to be acceptable-and meet the specification-are also reworked if the operator believes he can improve the appearance. Operators do not understand the reliability consequences of touching components with a soldering iron. They do know that their customers’ auditors often reject connections that are acceptable but not cosmetically perfect. No one gets disciplined for rejecting and reworking acceptable connections, whereas rebukes for passing a reliable but less than cosmetically perfect connection are very common.
Zero Defects was a hit with the military. The U.S. Army Missile Command presented a series of workshops and seminars about how companies could achieve ZD. Not to be outdone, the Department of Defense held seminars on the same topic. In 1964, Time magazine reported that 200 “major” U.S. companies had “adopted” ZD. But, as is so often the case by the time the mainstream press catches up with a management fad, early users of the approach were already defecting, and ZD was effectively dead before the end of the decade.
In a 1966 paper titled “Quality Problems, Remedies and Nostrums,” the dean of quality experts, J.M. Juran, lashed out at the Zero Defects mindset: “In all essential respects, the effectiveness of the ZD movement is grossly exaggerated; the unsuccessful programs have been more numerous than the successful; motivational programs have a narrow, not a broad range of application; the premises underlying the ZD programs are suspect; the main purpose behind the movement has probably been customer relations, not quality improvement.”
In modern quality terminology, ZD was quality control carried out at a fanatical level. And the modern quality professionals working in the electronics manufacturing industry are almost always quick to point out that quality control has long been supplanted by quality assurance. Quality assurance is supposed to bring about defect-free output the first time, before any inspection and certainly without rework. (Philip B. Crosby, the Martin quality engineer who ran the ZD program, would resurface in 1978 as the reinvented guru of “Quality is Free”-this time promoting the necessity of doing the job right the first time.)
IPC-A-610 and IPC-J-STD-001 are about classic quality control, about finding and reworking conditions that do not conform to the largely visual requirements laid out in their pages. No distinction is made between achieving the requirements through rigorous process management without rework and arriving at the same visuals with touchup and rework. In the IPC world, still mired in Zero Defects ideology, massive inspection and rework has as much virtue as perfect output without inspection and rework.
Critics of rework usually focus on labor content because rework (and inspection, for that matter) inflates labor costs. But there’s more to the story than higher labor costs. In particular, rework in electronics assembly plays havoc with reliability. Electronics can easily be damaged by mishandling (stacking units, for example), mechanical stresses from test, poor flux chemistry, static electricity and more. The primary killer of electronics reliability, however, is almost completely unknown because it occurs inside the component where it can’t be seen. The killer is the soldering iron.
Some types of heat damage-lifted pads, delaminated circuit boards, and melted component bodies to name a few-are easily recognized. However, soldering iron heat causes serious degradation inside components such as ICs where the damage can’t be seen. The most prominent example of such damage is accelerated growth of the intermetallic (“purple plague”) between the gold wire bond and the aluminum pad on the chip substrate. As the intermetallic grows, electrical resistance inside the connection increases and switching characteristics change; depending on the sensitivity of the circuit, this change alone can be fatal. Even worse, Kirkendall voids develop in place of the pad material and breaks develop around the edges of the pad. (For more about how soldering iron heat damages IC bonds, click here.
Heat kills reliability but it is also the reason why touchup can produce an apparently reliable connection on an unsolderable part. At machine soldering temperatures, solder will flow only if the area being soldered is clean and has been deoxidized; if the flux cannot remove the oxide (i.e., the part is unsolderable), solder will not cover the surface. Soldering irons, on the other hand, operate at much higher temperatures than machines such as wave solder machines or reflow ovens. The difference in temperature between machines and soldering irons is typically more than 120°C. And at those elevated temperatures, solder will adhere to oxides and contaminants on the surface being soldered. The solder appears to have flowed but there is no wetting and no intermetallic bond is created. A cosmetically perfect but functionally flawed connection can be made in this manner. In fact, soldering operators do exactly this hour after hour, day after day, making everyone from the company’s management to unwitting customers happy.
Although solder will bond to an oxidized surface at high temperatures, the time during which the iron must be in contact with the part is much greater than would be the case if the surface was clean and had been properly deoxidized. Consequently, touchup not only subjects the part to a second round of heat, the temperature is much higher and the contact time much greater.
In a sense, touchup and rework are all about deceiving the customer who, unwittingly, receives product with higher probability of premature failure. The military soldering standards (initially MIL-STD-454, then DOD-STD-2000 and MIL-STD-2000) on which IPC-J-STD-001 and IPC-A-610 are based were the results of negotiation between the military and their suppliers. Specifically, the defense contractors were looking for rules about how much less than perfect the military customers would be required to accept. The customer was obliged to accept any unit that met the minimum acceptable conditions, even if the results were less than perfect because of deficiencies in the contractor’s processes. Outside the military, commercial contract assemblers use IPC-A-610 and IPC-J-STD-001 in the same way, as licenses to ship reworked defects.
Ironically, many manufacturers are forced into adopting IPC-A-610 and IPC-J-STD-001 by their customers. Other manufacturers put their employees through IPC training in the belief that certification comforts customers. This is particularly true for suppliers in the defense and aerospace industries. I have experienced cases of customers rejecting output that, without rework, met the level defined by IPC as “Target” (what normal people would call “perfect”) because the operators and inspectors were not IPC certified. This does not happen in other industries. And it should not happen in electronics manufacturing. Rational customers should be looking for suppliers on the basis of proven output reliability, not ability to put a pretty cover on a homely connection.
The fundamental truth of electronics manufacturing is that reliability varies inversely with the amount of handling. All handling-even test-degrades reliability. But some handling is worse than others, and soldering iron heat is worst of all, far worse even than electrostatic discharge. Because of the rework mentality, costs go up and the product becomes less reliable.
We are now 30 years into the quality revolution launched by Juran, Deming and Crosby. But something has gone terribly wrong. Instead of the focus on results emphasized by Deming and Juran, industry has embraced paperwork bureaucracy. No longer manufacturers themselves, companies have lost the ability to assess their suppliers’ processes. Where a customer quality engineer or purchasing agent could once walk through a supplier plant and know the reliability of the processes, purchasing departments now depend on trappings such as ISO certifications. IPC-A-610 and IPC-J-STD-001 certifications fall into this category.
The tragedy of IPC-A-610 and IPC-J-STD-001 is resource allocation distortion. Companies operate with limited education budgets. Money put into one training scheme is not available for other learning initiatives. Rather than learning the science that delivers defect-free output without touchup and rework and providing that knowledge to their assembly personnel, electronics manufacturers spend the money on memory drills to identify defects. But, divorced from knowledge of the forces that determine output, the drills cannot produce comprehension of what process failures the defects represent-nor the corrective actions that will prevent recurrence of the defect.
But, the IPC backers claim, these are “industry standards.” “Standards allow manufacturers, customers, and suppliers to understand one another better. Standards allow manufacturers greater efficiencies when they can set up their processes to meet industry standards, allowing them to offer their customers lower costs,” is the statement at the beginning of both IPC-A-610 and IPC-J-STD-001. But this is not the case. Neither document provides meaningful process guidance.
For that matter, there is no reason to trust “industry” standards. IPC’s “The Principles of Standardization” found at the front of these documents says that “Standards should not… Contain anything that cannot be defended with data.” However, I have never been able to unearth any data that supports the criteria; everything seems to be based on opinion and compromise. Inertia dominates; many criteria were established decades ago for reasons lost in the mists of time and never challenged. Too many criteria are simply arbitrary; the default requirement for Class 3 products, for example, is typically 75 percent of perfect, while Class 2 product requires only 50 percent of perfect. This is dogma, not science.
Perfect soldering, whether by hand or machine, is not difficult. All it takes is a firm grasp of scientific principles like surface tension, other wetting forces, solderability management, fusibility and flux. This is the same knowledge customers require to evaluate suppliers of electronic modules. Above all, it is knowledge that must be provided to operators and inspectors, the people who are best positioned to identify and implement meaningful corrective action if a defect should occur. And it is not knowledge provided by IPC-A-610 or IPC-J-STD-001.
Inefficiency and compromised reliability may be compatible with profitability in a world where labor is close to free and products are not expected to last beyond a couple of years. However, that world is not the environment in which Western manufacturers operate. Our labor is far from free, and our products are expected to be more durable. Survival here depends on doing the job right the first time, every time. Training people to expect rather than prevent defects invites misfortune. It’s time to learn how to do the job right the first time.
Editor’s note: Before “Shipulski on Design,” “Leading Lean,” and “Uncommon Sense,” there was ASSEMBLY magazine’s longest running and most controversial back-of-the-book column, “Unconventional Wisdom” by Jim Smith. A nationally known expert on electronics assembly, Smith never hesitates to question the sacred cows of manufacturing and economics. You can read more from him at his “Science of Soldering” blog http://blog.emsciences.com.