The client had an odd problem: a type of chip capacitor was jumping about an inch to the side during reflow. It didn’t happen with every capacitor, but it did happen with many. Why did this happen?

At first, it was only one client. But soon I heard the same complaint from several other clients. And many clients were noticing increased numbers of tombstoned capacitors. All of the clients wanted to know what was wrong with their soldering processes.

There was nothing wrong with their soldering process. They simply bought bad parts.

Whenever a defect involves a particular component, the first troubleshooting step is to see what is different about that component. In this case, the capacitor was manufactured by a low cost Asian supplier and sold for less than the similar component from more established manufacturers. As is so often the case, the lower price was more than offset by the costs of rework.

Ceramic chip capacitors consist of layers of ceramic sheets printed with conductive ink. The layers are fused together under pressure to form a laminate which is fired in a kiln. The ends of the laminate are then dipped in a mixture of metal powder, glass frit and solvent to form the terminations. The component is fired in a kiln again to fuse the metal/frit mixture to the ceramic laminate. (A similar process is used to apply the resistance heating strips of automobile rear window defrosters.) This is where the jumping capacitor problem originates.

The fired metal/frit termination is somewhat porous. Achieving small pores requires much better process control (and higher cost) than the steps that result in larger pores. To cut costs, the component manufacturer did not exercise adequate control over the termination processes and got very porous metal/frit as a result.

After the metal/frit has been fired, the component goes to an electroplating bath for application of nickel to the terminations. The plating bath contains salts that can get trapped under the nickel if the pores of the metal/frit termination are too large.

During reflow of assemblies with those capacitors, the plating salts vaporize and the termination explodes. The force of the explosion can be great enough to cause the component to jump off the pad. Less vigorous outgassing causes tombstoning (the component is tipped so that one end is not soldered).

Outgassing is almost unknown with components from the more established manufacturers but is being seen more often as companies buy from second-tier suppliers that offer lower prices. But the lower price can be very expensive if the components shift during reflow.

Robotics and automation can cut production costs significantly. But too many companies automate steps that were not necessary in the first place – and that’s just flushing money down the drain. I have a few rules about automation and robotics that I tell my clients. You can read about them in detail at Assembly Magazine’s web site http://www.assemblymag.com/blogs/14-assembly-blog/post/90711-working-with-the-lights-out

Three of the rules are:

1. Recognize that human beings are more flexible than machines. If products or components change, it may not be possible for a machine to adapt. People, on the other hand, can probably make the adjustment quickly and inexpensively.
2. Be sure the activity to be automated is really necessary. I have been in many, many plants where costly process steps could have easily been eliminated. If the steps were being performed by humans, eliminating the unnecessary procedures would have saved lots of money through wage reductions. But once the steps have been automated, there are no wage reductions available by being more efficient.
3. Use realistic amortization schedules in assessing the potential R.O.I. that a new piece of equipment would provide. A machine that might pay for itself in five years is just money wasted if the machine only provides three years of operation.

These rules are pretty straightforward. But too many companies don’t take them into account when contemplating whether to reduce workers with machines. The consequences can be fatal.

You can read more at Assembly.

Much of my work this year has involved solving flux problems. And most of those problems have involved organic (OR) fluxes, including those with the lowest acidity that do not contain halides such as chlorides. The evidence is clear — organic fluxes are not compatible with reliability, especially for assemblies that operate in humid environments.

I’ll explain one problem with OR fluxes in a moment. But, first, a few words about what is meant by “organic” may be helpful…

The primary function of flux is oxide removal; solder can’t bond to an oxidized surface. But eliminating oxides is not enough since the deoxidized surface will immediately reoxidize if oxygen can reach the surface. So fluxes also contain materials, generally called “solids,” that form a barrier between the deoxidized surface and oxygen.

Traditionally, electronics fluxes relied on rosin to seal the surface. It’s basically the same as the rosin musicians put on violin bows and baseball pitchers use to better grip the ball. Rosin works especially well for two reasons:
1. It withstands soldering temperatures, and
2. It is hygrophobic, forming a barrier against moisture that acts like conformal coating after soldering. This property is especially important because the residues of flux acid after soldering can be electrically conductive and corrosive, but only if moisture is present.

But rosin has some characteristics that are less lovable. In particular, rosin is sticky, easy to see after soldering and ugly. Customers tend to believe that rosin residues on the circuit board mean the system is likely to fail, so electronics manufacturers feel compelled to remove the residues. And rosin removal is very expensive.

Replacing the rosin with substances such as glycols results in circuit assemblies that are close to invisible without cleaning. (There are ways to use rosin fluxes that produce circuit assemblies that look clean without actually being washed after soldering. I will discuss one approach in a future blog.) These non-rosin fluxes are termed “organic” (OR) even though rosin itself is an organic substance. (The terminology is complicated and confusing but I will explain it out in a future blog.)

But glycols and related chemicals, unlike rosin, attract moisture. In chemical jargon, they are “hygroscopic.” And this can have catastrophic consequences.

The images below are the results of using ORL0 and ROL0 on the same circuit assemblies without cleaning. The capacitors of Image 1 were soldered with ORL0 flux (the weakest of the OR fluxes) then the system was powered up and subjected to 95% relative humidity. The capacitor of Image 2 was soldered with ROL0 (the weakest of the RO fluxes) and subjected to the same humidity when the system was powered up. The ORL0 flux caused catastrophic corrosion and dendrites that did not occur with ROL0 flux.

Let me be clear. Everything about Images 1 and 2 is identical except the flux. ORL0 is the weakest organic flux, nominally the same as ROL0 flux. Rosin provided ample protection from humidity. The organic flux residues attracted moisture that caused serious failures.

I repeat — organic fluxes are not compatible with reliability, especially for assemblies that operate in humid environments.

A few months ago, I wrote some blog entries about the benefits of nickel barriers between the base metal and surface plating on component leads. Basically, the nickel has much less tendency to react with other metals than do tin, copper or steel, so a thin nickel barrier slows the rate of intermetallic growth. Reducing the intermetallic formation provides two major benefits:

1. Intermetallic takes much longer to reach the surface. This is significant because intermetallic oxide cannot be removed by fluxes safe for electronics use.
2. There is evidence that tin whisker formation increases as the intermetallic layer grows. It seems that the increasing intermetallic produces physical stresses on the tin layer, causing the sprouting of whiskers.

Justin Kumpf, a reader of this blog, posed the following questions:

1.  If a part has had a nickel barrier coat applied below the tin, but the inter-metallic layer has broken through, is there a remedy? I was thinking another layer of nickel and then more tin? Will it be possible to plate over the IMCs? Our plater tried to strip both the nickel and tin off of some of these parts, but his process doesn’t seem to be removing the nickel completely. Is there a process that will get us back to base metal?
2. He (the plater) is just stripping with a nitric acid bath, is there an electrolytic solution?
3. Oh, the base metal is mu-metal, so that complicates matters I expect?

Provided the rest of the component can withstand the lead restoration processing, Justin is closer to the solution than he seems to realize. It is possible to plate over the intermetallic. (For that matter, it is possible to plate over a twig, a credit card, even a Q-tip. Car “chrome” is typically deposited over a plastic substrate.) But is that a good idea?

Intermetallics have a couple of properties that do not match up well with electronic reliability. First, they are poor conductors. Second, they are brittle and prone to cracking after repeated thermal cycling or vibration. So, simply plating over the intermetallic may not produce a happy result. (There’s an oxide layer over the intermetallic where it has broken through to the surface. Again, it is possible to plate over the oxide. But oxides are insulators and plating over the oxidized intermetallic will compound the resistance issues raised by the intermetallic alone. Therefore, at a minimum, removing the oxide before plating would be good practice.)

Although removing some of the intermetallic would be desirable, there’s no need to get back to base metal before replating. Plating over a thin intermetallic produces a part that would perform very much like the original. For that matter, there’s no need to apply a new nickel barrier layer unless the time between replating and soldering will be extended. (If the intermetallic oxide is not removed before replating, however, the tin will melt during soldering and dewetting will occur.)

Nitric acid should have sufficient strength to remove the intermetallic as well as the intermetallic oxide. However, the formulation used in electroplating tends to be relatively weak and very long immersion (hours, potentially) could be necessary to completely etch the surface back to base metal. The only way to know the time required is by testing. Using less diluted nitric acid would accelerate the etching.

A mu-metal base doesn’t really complicate matters. The resulting intermetallic between mu-metal and tin will not be thick but that is actually a positive outcome (recalling that thicker intermetallics increase electrical resistance and can crack).

---

 
I welcome questions about soldering. You can post questions using the comments box below.

The last two entries of this blog answered questions posted by reader Binh Pham. One final question remains:

If a product is required to be complaint to class 3, but the heel fillet of one gullwing lead does not meet the A-610 class 3, what are the reliability implications to the product if I do nothing? Is life testing required to understand the impact?

Class 3 specifies that the top of the heel fillet be level with or above the top of the lead foot. (Class 2 settles for the heel fillet reaching just half way up the lead foot.) Is that a good standard? As I explained last time, I have never found any data backing up any of the criteria. That doesn’t necessarily mean the criteria are defective (though I take exception to quite a few) but it does mean there’s no way to know the reliability consequences of failing to reach the Class 3 minimum. Also, Pham doesn’t specify the extent to which the heel fillet fails to meet the Class 3 requirement or whether the deficiency exists on just one or many leads.

However, Pham’s question indirectly gets at what I see as the fundamental fallacy of the “one size fits all” standard – that there are great differences among products and components. The component mass and number of leads are two critical variables, but so are type of laminate, number of layers, metal content and all the other factors that determine the PCB’s coefficient of thermal expansion. Greater stresses will be placed on the solder joint as the CTE mismatch between component and PCB increases. Also, what will be product’s operating environment?

A standard should be a guide, not a commandment, and the manufacturer should have the option of demonstrating reliability through appropriate life cycle testing. But it must also be recognized that life cycle tests themselves can be prone to faulty results.

On the other hand, there is nothing too onerous in most Class 3 solder requirements. After all, most of the specifications say only that the solder connection should be not less than 75% of “perfect.” That’s usually not hard to achieve with proper process management.

Ultimately, however, the reliability is probably irrelevant because the Class 3 requirement is normally spelled out in the contract with the customer. The contract is legally binding and trumps all other considerations. There are circumstances where customers may waive the contractual requirement, but they tend to be rare. Or, perhaps, the manufacturer has no customer requirement and is setting the requirements itself, but that, too, is unusual.

So, what’s the bottom line? By contract, Pham is required to rework (touchup) the deficient solder connection, subjecting the component to additional thermal stresses and degrading reliability in the process. It’s not a rational act, but that’s the nature of the Class 3 world – the appearance of reliability takes precedence over the real thing.

In my last post, I began responding to questions about Class 3 criteria of A-610 posed by reader Binh Pham. His first question – How did the industry association that writes the standards come up with the requirements? – was addressed last time. This time, the topic is his second question – Does the association have field and/or life data to support the requirements?

The short answer is: no. Or, more accurately, not really.

As I explained in the last article, A-610 grew out of MIL-STD-454. And MIL-STD-454 was the result of negotiation between defense contractors and the Department of Defense. The contractors wanted the least restrictive requirements while the DoD wanted nothing but perfect solder connections. The result (which, generally, says that 75% of “perfect” soldering is acceptable for DoD purposes) was the product of prolonged debate and not a small amount of lobbying by the contractors. Science and data played no meaningful role in the negotiations; decisions were based on hunches and emotion. (For that matter, little was known about the science of soldering at the time the decisions were being made.)

I have some first-hand knowledge of how solder requirements were established because I was starting in electronics assembly at about the same time. The first circuit boards had appeared not too long before and wave soldering was emerging as state of the art for automated soldering. The circuit boards were single-sided; through-holes were not plated.

Solder connections of single-sided circuit boards often contain voids, the result of surface tension pulling the solder to the point where the lead is closest to the edge of the pad. These openings caused much passionate debate about how much, if any, void would be allowed. Some people refused to accept any gap. Others allowed voids that were not greater than 25% of the through-hole area. Some even considered 50% gaps acceptable. But there was no science behind these opinions, only opinion.

The same debate about hole fill arose in new form with the arrival of double-sided circuit boards. This time, the controversy concerned how much vertical through-hole barrel fill was necessary. Once again, there were groups who required 100% vertical fill, others willing to settle for 75% and still others who accepted 50% fill. (The not inconsequential question of how to measure the amount of fill less than 100% was conveniently ignored, as continues to be the case.) This is how standards were created.

A-610 is full of criteria based on such subjectivity. All of the criteria that express a requirement in percentage terms (50% or 75% of a surface mount lead on a pad, for example) are completely arbitrary. There is no science behind the requirement.

Would “data” be meaningful even if it did exist? Probably not. The reliability of anything less than perfect will be a function of many variables such as component mass, PCB features (how many layers, type of laminate, metal content, just to name a few), soldering process, total mass of the solder connection, number of leads, mechanical supports, and much more.

Next time, I will look at Binh Pham’s third question about Class 3 criteria – If a product is required to be complaint to class 3, but the heel fillet of one gullwing lead does not meet the A-610 class 3, what are the reliability implications to the product if I do nothing? Is life testing required to understand the impact?

Reader Binh Pham has written to ask the following set of questions regarding Class 3 criteria of A-610:

1. How did the industry trade association come up with the requirements?
2. Does the association have field and/or life data to support the requirements?
3. If a product is required to be complaint to class 3, but the heel fillet of one gullwing lead does not meet the A-610 class 3, what are the reliability implications to the product if I do nothing? Is life testing required to understand the impact?

The foundations of A-610 are found in U.S. Department of Defense (DoD) standards, especially MIL-STD-454, first issued in the mid-1960s. Defense contractors negotiated with the DoD to set out formal rules specifying the extent to which solder connections could deviate from perfect but still be acceptable to the government. If solder connections met the rules even though they were not cosmetically perfect, the government would buy the product.

The original soldering specifications occupied only 7 pages (including large illustrations) in MIL-STD-454, but the number of specifications kept growing. 20 years later, the soldering section had expanded to 25 pages. DOD-STD-2000, which replaced MIL-STD-454 in 1986, consisted of three high reliability (corresponding to today’s Class 3) books of roughly 75 pages each and a fourth book of requirements for what the DoD termed “general purpose” soldering (probably equivalent to today’s Class 2). Some of the added content related to requirements for tools, equipment and facilities, but the soldering requirements were much more detailed than had been the original case. MIL-STD-2000, which combined and replaced the separate books of DOD-STD-2000 in 1989, ran to more than 200 pages.

The first edition of A-610 was published in 1983. Most of the document reflected MIL-STD-454 and Class 3 is entirely DoD-based. In 2002, the DoD simply turned standards development over to the trade association and A-610 has been the military’s unofficial reference standard ever since.

Next time, I will look at the question of data supporting the criteria.

Reuters has an incredible, depressing report that GM refuses to honor suspension warranties on 2007 and 2008 Impalas (http://www.reuters.com/article/2011/08/19/gm-impala-lawsuit-idUSN1E77I0Z820110819). It seems that, despite frequent and very loud claims to the contrary, America’s corporate heads still don’t understand quality and consumer perceptions. The story has eerie and troubling overtones of the 1970′s quality crisis that almost destroyed the U.S. auto industry and opened the door to brutal foreign competition.

Reuters says some 400,000 Impalas have faulty rear spindle rods (no, don’t ask me to explain what they are – I’m just a soldering guy – but apparently they are very important) that can cause exceptionally rapid rear tires wear. A class action suit seeking damages claims that the rear tires of an Impala owned by Donna Trusky of Pennsylvania wore out after 6,000 miles because of faulty spindle rods. The suit claims that GM replaced the rods on police vehicles but not cars of private owners.

GM says it isn’t responsible because the entity that the vehicles were manufactured and warranties were issued by “old GM” that went into bankruptcy in 2009 and is now legally known as “Motors Liquidation Co.” The GM now making and selling vehicles styles itself “New GM” and accepts no responsibility for anything done by Old Good Morning.

It turns out this isn’t the first time “new” GM has hidden behind the “old” GM entity. Reuters reports that an earlier suit concerning problems with OnStar in pre-bankruptcy vehicles was dismissed on the same grounds alleged here.

Of course, companies go broke all the time and use bankruptcy to escape responsibility for past acts. But GM is (we thought) different. After all, U.S. taxpayers collectively own more than 25% of “new” Good Morning, and some of those vehicles were sold after President Obama told prospective GM buyers before the bankruptcy:

“Let me say this as plainly as I can. If you buy a car from Chrysler or General Motors, you will be able to get your car serviced and repaired just like always,” Obama said in a speech. “Your warranty will be safe. In fact, it will be safer than it has ever been. Because starting today, the United States will stand behind your warranty.” (http://www.autoweek.com/article/20090330/CARNEWS/903309977 )

More fundamentally, vehicles are the biggest purchase after a house that most consumers ever make. Will they trust a company now, after seeing it hide behind a legal nicety? Will they buy GM when Ford hasn’t turned to courts for protection? For that matter, will they even buy American?

It took more than 20 years for Detroit automakers to dig out from the rubble of the 1970’s rust scandal. This story has the potential to set the industry back another 20 years. Ultimately, I expect the federal government will end up paying these claims and the public will (rightly) be very angry.

American companies have been talking a great quality line in recent years. But actions show that, for many of those companies, the claims have no substance. In electronics, major U.S. corporations have closed down there in-house manufacturing and outsource production to cheap labor countries where cost always trumps reliability. Even companies that continue to produce here endorse rework and touchup as valid ways of meeting the visual requirements of A-610 and J-STD-001, knowing perfectly well that a reworked connection translates into premature failures.

Also this past week, HP pulled the plug on its TouchPad tablet, only six weeks after bringing it to market. TouchPad owners – people who trusted HP’s claim that there would be ongoing support – are out of luck. There will be no software updates, no well-stocked apps store. Most retailers are refunding the purchase price if owners return their TouchPads, but many of those buyers probably invested considerable time customizing the programs and getting familiar with the operating system; a refund doesn’t restore their lost time and it certainly won’t make those people line up to buy HP’s Next Great Thing. (Many people did line up for TouchPads when HP slashed the price by 75%, so many that HP’s computer service infrastructure crashed under the weight of the buying frenzy. Now HP wants to concentrate on business IT, but what IT manager is going to trust HP to run the manager’s bits & bytes when HP couldn’t keep its own system running.)

Think about it: this is HP, onetime star of America’s technical equipment universe, not Joe’s Garage and Computer Outlet. But it’s also the HP equivalent of “new” GM; it wants out of the consumer market and seems determined to burn those consumer bridges. It’s the world’s largest computer equipment company and it wants out of the computer business! Who do we trust after such betrayal?

America has fallen into the trap of confusing bureaucracy and paperwork (think much ISO activity and not a little of “certifying” employees who memorize but do not understand the meaning of A-610 and J-STD-001 requirements). “Supplier quality” management has deteriorated to mindless comparison of ISO paperwork to actual process. Only occasionally (and, I sometimes think, only by happenstance) do the supplier quality people actually get their hands dirty with production people.

Right now, America is getting by with this sorry excuse for “quality” systems. But, sooner or later, one of those developing countries currently noted for providing American companies with cheap but shoddy products, will undergo a transformation like Japan in the 1970’s. (And whatever happened to the legendary Japanese quality? Many Japanese companies have been sending work to underdeveloped Asian countries, much like American companies, with the corresponding fall in reliability.)

Western companies face enough problems trying to compete with dirt-cheap foreign competitors. The only hope for survival is to make and stand behind better products. The news from GM and HP isn’t promising.

Last time, I wrote about the key fob solderability fiasco. A key fob is the key ring gadget that remotely locks and unlocks car doors. The fiasco involved an unsolderable critical component in the key fob that the manufacturer “fixed” by adding more solder at high temperature. The “fix” produced connections that met the visual requirements of A-610 and J-STD-001 but only disguised the fatal defect. Hundreds of thousands of key fobs with defective solder are failing in very large numbers because of the manufacturer’s lack of knowledge about solderability.

Sadly, the key fob is not the only product failing in large numbers because manufacturers do not understand solderability. Solderability management is the single greatest challenge faced by the electronics industry today because a ban on lead in electronics by Europe and a handful of smaller regions. Solderability is the lead–free challenge killing electronics reliability; other lead–free issues such as higher melting temperature and lesser wetting action are trivial by comparison.

Solderability is poorly understood because the electronics industry has rarely soldered in the past. Most of what was called “soldering” was actually welding. That is, the surfaces – almost always plated with tin or tin/lead – being “soldered” melted during application of the solder. When surfaces melt, the process is welding, not soldering. Soldering, on the other hand, creates intermetallic bonds between solder and a metal surface that does not melt.

Unable to sell components with tin/lead plating in Europe, component manufacturers have largely eliminated tin/lead everywhere. They could continue to produce tin/lead-plated parts for areas like North America but it is easier and cheaper for them to offer just one version. Simultaneously, concern about tin whiskers is eliminating tin plating. The new component surfaces do not melt at soldering temperatures. So, after decades welding tin and tin/lead parts, electronics manufacturers suddenly must truly solder. And they don’t know how.

Unwittingly, the electronics assembly industry has long employed terminology showing that someone, at some time, recognized the true nature of the “soldering” operation. The term is “reflowing” (as in “to melt again”) and it remains in common usage today, although it is being applied to the new lead–free components with surfaces that do not melt.

More significant, the traditional welding processes are still being used even though the surfaces don’t melt. A welding process will cause wetting defects if the surfaces do not melt (although a soldering process, being more robust, can be used in welding).

Metallically bonding a surface that melts (i.e., welding) is not very challenging. Oxides and contaminants – the materials that can prevent intermetallic bonding – are light and float on the surface of the liquid solder. Like the oxides, the flux floats on the liquid surface and has no trouble reaching the oxides. Moreover, the oxides of those melted surfaces – tin and lead oxides – can easily be broken down by even the weakest fluxes such as Type R flux that consists of only rosin in alcohol. So, “soldering” surfaces that melt is about as hard as breathing.

Soldering has two requirements that don’t exist when surfaces melt:

First, the flux must get to the surface and remove the oxides before the solder melts. In hand soldering, this means applying liquid flux before soldering; flux-cored solder doesn’t release the flux until the solder melts, after which the dense solder blocks the flow of the (light) flux.

Second, the flux must be able to deoxidize the surface. That was not a problem with tin or lead, both of which form weak bonds with oxygen. For that matter, even lightly oxidized copper can be deoxidized by a mild flux. But the new surfaces cannot always be deoxidized with fluxes compatible with electronics reliability.

Until recently, electronics manufacturers could pretty much ignore solderability – and they did. They could get away with using a welding process and that made them careless. But solderability can no longer be taken for granted and processes based on surfaces that melt must be replaced. Key fobs show us why.

Solderability problems figure prominently in A-610 and J-STD-001 defects. But, “nonwetting” aside, they are not always presented as solderability defects. For example, “cold” solder (a condition that, as has been pointed out here from time to time, has nothing to do with lack of heat) is not presented in solderability terms, although it generally reflects either solderability or surface contamination. “Insufficient solder” in the form of “wetted fillet is not evident” presented in Figure 8.36 of A-610 Rev. E is mostly about solderability, too. And solderability is one common cause of incomplete vertical fill of a plated hole.

Unsolderable surfaces always produce defects in wave soldering or surface mount reflow because solder will not flow in these machine processes unless all oxides (and any contaminants) have been removed. Hand soldering is different and can produce cosmetically acceptable connections on non-wetted surfaces; the temperature of hand soldering is much higher, allowing solder to stick to oxides (without intermetallic bonding), and the operator can push the solder into shape. If the surface was not solderable, touchup/rework only covers up the problem. (If both hand and machine soldering use compatible fluxes, as J-STD-001 3.3.1 requires, the hands soldering flux cannot deoxidize a surface that was not deoxidized by the machine flux).

In other words, the touched up/reworked connection is still defective (albeit more attractive) but it conforms to the J-STD-001 and A-610 requirements! This absurdity happens countless times all over the world every day. I have said it before, but it deserves repeating: a hand soldered connection that meets the cosmetic acceptance requirements of A-610 or J-STD-001 is not necessarily reliable. The criteria only have relevance to machine soldered connections that have not been touched by a soldering iron.

Since touchup/rework is allowed and works as camouflage, companies fail to recognize hugely important solderability problems. I was reminded of this truth again today when the cable repairman asked me about a problem he was having with his car’s remote for locking and unlocking the doors. Auto companies call these gadgets key fobs. Both of the key fobs for his 2008 vehicle had failed because the metal strap that normally holds the battery in place separated from the solder. “Can you fix these for me?” he asked.

As it happened, I am very familiar with the key fob in question. Shortly before the great economic meltdown, during a soldering engineering workshop I was conducting at the auto company in question, a supplier quality engineer showed a sample to me. The engineer was concerned because the metal strap that retains the battery had failed during validation testing. “Of course it fails,” I replied. “It isn’t solderable. The solder is sticking to the surface oxides but there’s no intermetallic bond and no mechanical strength.”

A few weeks later, I asked the engineer how the supplier had resolved the problem. “They’re adding more solder,” he replied. And that is the way so many solderability problems are “fixed.” Unfortunately, the fix only kicks the reliability can a bit farther down the road and, before long, the real bill comes due.

Adding more solder doesn’t cure non-wetting; it only masks the problem for a while. And, as my cable repairman discovered, a disguised reliability problem is still an incipient failure. At this moment, there are undoubtedly tens of thousands of failed key fobs with unhappy owners.

Disgracefully, this key fob disaster is not unique. In fact, it is not all that rare. It happens because too few people – soldering authorities included – understand solderability.

The next blog entry will explain why solderability management has become far more important in recent years.