Pennies saved on the front end can cost big money in rework.

Ever since I began assembling PWBs nearly 20 years ago, I’ve been addressing problems whose root causes lie not within assembly but PWB fabrication. I did always appreciate that fact. Then, as luck would have it, I transferred under a boss whose background was in fabrication. He opened my eyes to the influence of the fabrication process on the assembly process, and how to investigate, prevent or compensate for the bare board’s effects.

As we gain Pb-free experience, we learn about the impacts of the “upgrades” both to fabrication and assembly. As it turns out, just like every other major factor we’ve studied during this transition, influential considerations in SnPb processes are just as influential in Pb-free, which always has a few extra-special concerns. I have yet to encounter a SnPb process consideration whose Pb-free corollary doesn’t have some distinctive twist.

Some of the general fabrication issues we have learned to anticipate regardless of alloy include:

• PWB feature location accuracy. Sometimes, locations of the centroids of the devices on the PWB do not match their CAD file locations. The lamination process often causes PWBs to shrink from their nominal locations. To compensate, fabricators typically factor the expected shrinkage into their imaging processes, but variations in raw materials and processes can push the “fudge factor” a little off the mark. The result for the assembler? Misaligned paste prints and misplaced devices, despite good process control on the stencil fabrication, stencil alignment and pick-and-place. The paste deposit and placement might be exactly as designed from an X-Y perspective, but if the landing pads are off-target, they can undermine many of the benefits running good process controls on the assembly line.
  
  
• PWB feature size accuracy. Copper overetch and underetch on the PWB surface can cause problems on the assembly line. Overetch seems to be the greater menace, as it renders pads slightly (say 0.0005" to 0.001") smaller than they should be. Smaller-than-expected pads can cause gasketing problems during printing, resulting in bridges and random solder balls. A slight bit of overetch can cause wet bridges in the paste print, and the root cause can be very difficult to pick up with an unaided eye. Over the years, I’ve witnessed some great engineers stymied by undiagnosed overetch issues, spinning their wheels investigating paste viscosities, print parameters, support tooling – all to no avail.
  
  
• PTH drilling and plating. Poor drilling or desmear can leave glass fibers protruding into the PTHs that breach the plating coverage. The tiny breaches provide paths for internal moisture to outgas during soldering, resulting in blowholes. Typical SnPb blowholes look like mini volcanoes, with a small pinhole at the top, and can usually be addressed by baking moisture out of the PWB prior to soldering. Pb-free blowholes are one of those special considerations – those distinctive twists – and are not so easily resolved.
  
  
• Final finish. I learned the hard way that not all final finishes are created equal, and if you don’t specify exactly what you want, you get what the fabricator chooses. That can mean the lowest cost option for the fabricator, which is not necessarily the best performance option for the assembler. It’s not unreasonable to anticipate a final finish’s solderability and shelf life, and thermal degradation resistance will take a backseat to the bottom line if specific product or performance criteria are not identified upfront.
  
  
• Solder mask selection and application. Just like final finishes, not all solder masks are created equal, and if you are not specific in your selection, you get “whatever.” If the mask du jour is not the least expensive material available, it is likely whatever was set up on the line from the prior run, which means it may vary with each lot of boards. The importance of mask selection on soldering performance should not be underestimated. Glossy masks can hinder wave soldering by limiting a liquid flux’s ability to spread evenly across the PWB surface, and by creating micro solder balls (MSB) that stick to the board. Undercured masks of any variety can also contribute to MSB adhesion, and even microdross adhesion. I recall one particular batch of PWBs that passed visual inspection after soldering, but came up full of shorts at ICT. Closer inspection revealed microscopic pieces of dross were clinging to the mask, shorting out vias to pads and to other vias. If the ICT failure reports hadn’t told us the exact locations to examine, it may have taken a long time to find them visually. They were fine, dull, brownish-grey webs that clung to the mask. But they were conductive, and formed electrical shorts that could not be seen under the ring light given typical inspection conditions.

Some bonus concerns we need to watch for in Pb-free processing include:

• Copper erosion. This can be a minimal risk or a huge concern, depending on the PWB. We’ve still got a lot to learn about it. Essentially, the electrodeposited copper on the PWB dissolves quicker in Pb-free alloy than in SnPb, and soldering cycles with long dwell times can remove enough copper to affect signal integrity or interconnect reliability. In a recent test of 10 fabricators, the erosion rates of various ED coppers varied by a factor of almost 2X. As an industry, we don’t yet understand why one plating bath can produce a copper with more or less erodibility than another bath, but we have documented the differences, and the investigations continue.
  
  
• Blowholes. As I mentioned, this seems to be one of the phenomena that just gets bigger and badder in the Pb-free world. Pb-free boards seem to blow out more frequently than do SnPb ones; the resulting PTH voids are bigger than in SnPb, and those defects that used to look like pinholes in the otherwise normal SnPb solder fillets now resemble balloons – inflated radially, with a very thin shell of solder that can be broken just by poking it with a fingernail or small hand tool. Adding insult to injury, simple bake-outs do not seem to fully resolve the problems. We’ve documented several cases where a PWB produces blowholes on a Pb-free wave, but none on a SnPb wave, so copper erosion may play a part in blowhole formation as well. Again, we’ve got a lot to learn, and investigations continue.

It would be unfair to describe a number of potential problems without offering some practical preventions or remedies; I would expect to get my share of hate mail from the fabrication community if I fell short of offering some positive advice. I’ve learned a few tricks over the years to either prevent or work around fabrication issues on the assembly line, and I’ll be sharing them next month.

In the same vein, it would be unprofessional for me to make blanket generalizations that bash all board fabricators for lousy quality. Not all circuit cards carry issues that impact assembly yields. I’ve experienced an enormous performance spectrum in the market – especially during the transition – and some topnotch shops make great boards. It’s no wonder why they are often a little pricier than their competitors.

This month’s lesson learned is almost too cliché to even print, but it’s a truism we experience in many facets of our lives: You get what you pay for. In its quest to continually lower costs, our industry has attempted to commoditize many elements of the assembly process, judging them by acquisition cost alone rather than total cost of ownership. In many cases, we’ve forgotten saving a few pennies on the front end can cost us big dollars in rework at the backend. Sure, passive components or memory devices can be viewed easily as commodities, and changing suppliers typically has a negligible impact on an assembly operation. But not all constituents of the assembly process are interchangeable, and my experience with PWBs indicates they are one of the elements that should not be viewed as a commodity.

I am certainly no expert on the PWB fabrication process. Rather, I’m one of those people who knows just enough about it to be dangerous (and occasionally annoying to those who possess real expertise). But from the 30,000-ft. perspective, it does not substantially differ from assembly in the sense that successful quality requires an intricate balance of materials, equipment and process controls. That delicate balance is rooted in solid science, but much of it is achieved only through years of experience and sustained operations. It’s time to take PWBs off the commodity list, begin appreciating good fabricators, and be willing to pay for the quality we demand. The alternative: We will never get what we want, only what we deserve. And we’ll continue to pay too much for our PWBs through the hidden costs of rework, repair and returns.

Chrys Shea is an R&D applications engineering manager at Cookson Electronics (cooksonelectronics.com); chrysshea@cooksonelectronics.com. Her column appears monthly.