Despite the many technology advances in the SMT process, there will always be a need for rework and repair of PCB assemblies. Most especially as the industry trend continues towards finer pitch and spacing, smaller and smaller components, and increasing high-reliability requirements from customers.
Not to mention that fact that the rework and repair processes are already challenging. You have to deal with already-finished boards, wherein improper soldering may cause collateral damage to around the area that needs rework. So, for this month’s issue of SMT Magazine, we interviewed Dan Patten, general manager, and Laura Ripoli, customer service manager for rework/repair, from BEST Inc.; Gary Freedman, president of Colab Engineering; and Andy Price, sales manager of Circuit Technology Center, to know more about the rework and repair of PCB assemblies, the critical challenges, and strategies to improve the process moving forward.
“We’re always making more densely populated boards and we’re making things smaller and packing more into it and so forth,” says Patten.
Price notes, “From our perspective, some of the most challenging issues that we confront daily tie in with the circuit board design. The spacing is tighter, the boards are thicker and contain heavy copper planes. Those factors along with the RoHS requirements help create many of the challenging issues that we face in trying to rework components. It also increases the challenge in getting complete solder fill in plated holes. Much of it is directly related to the circuit board design. Some of the other factors include rework and repair on boards that have gone through final processing. They may be field returns or failures that are found after conformal coating application, or through the under-fill process at BGA component sites. There’s are many problems that can arise when reworking boards with conformal coating. Often collateral issues to adjacent components can occur if the area is not prepared properly such as starved solder joints, solder shorts and other problems. Those are just a few general thoughts on some rework issues we confront.”
Price notes that board thickness is a challenge. “Years ago, board thicknesses ranged from maybe 0.062” to 0.092”. Now, we’re routinely seeing boards that are 0.125”, and even thicker. There are ground planes involved in the construction of these boards that dissipate heat when you’re trying to remove components. Then you add in the fact that you’re dealing with a higher temperature reflow with RoHS requirements. These are just challenges that we have to deal with. I don’t see it going away. I see it becoming more and more challenging,” he explains. “The prevalent issues that we see with circuit boards that are constructed like that relate to the increased heating requirement. Boards require preheating and complete bake-out prior to rework. A lot more heat is needed in the rework area to achieve reflow temperatures. When more heat is necessary you increase the potential for physical damage to the board to occur. You may see baseboard discoloration or more significant base board damage such as delamination. Increased heat at the rework site can also result in lifted lands, surface mount pads and pulled plated through hole barrels. These are issues that anybody involved in rework will see, the most expert companies will see less of it due to experience but there are some designs that are very difficult to overcome.”
One of the problems that Freedman had seen very often was trying to repair things that don’t need repair on boards, such as barrel fill that doesn’t quite meet IPC spec. “One of the things that I had been driving for years when I was with Digital, Compaq and HP was trying to refine the barrel fill specifications. We have had some success in driving that movement with IPC. If you look at a through-hole solder joint, it’s many times stronger than an SMT solder joint. Yet, we’re happy to pass SMT solder joints with very little solder. When it comes to through-hole, people are so quick to take a soldering iron to it. It might just be the wrong thing to do,” he points out. “What we want to do is minimize the number of heat cycles on a circuit board. Besides the number of heat cycles, we also want to minimize the number of touches. Any time you touch a board, you’re likely to damage it, either through flexure or shearing of parts on the back side, or on the front side, with the soldering iron. Or, I see operators with picks all the time, poking and prodding to see if things are soldered. Like I said, anytime you can minimize the number of touches to a board, you’re better off. I’m one that is much happier to pass a board that doesn’t quite meet IPC than to have it go through a repair cycle. Chances are, the repaired board will be of a lesser reliability.
“A through-hole solder joint is many times stronger than a surface mount solder joint. We get very particular about these massively strong solder joints and yet we’re very willing to pass SMT solder joints that have almost no solder. I certainly understand that there are different mechanics involved in expansion of contraction, etc. We turn a blind eye to SMT, whereas with through-hole we get overly particular. I said that there are two issues: number one is the number of heat cycles we apply to a solder joint. That, in part, defines intermetallic characteristics and reliability. The other thing is the number of touches to a board. Any time a board is subjected to rework, you’re more apt to create damage than if the board is left alone.”
For years, Freedman had done a lot of work on reliability of solder joints, including through-hole solder joints, as well as a lot of modeling and testing of solder joints to revise their specification requirements and to also influence IPC.
“And we have done that to some degree,” he says. “If you have a very thin board, IPC’s very happy with a minimum amount of solder joint. If you have a thick board, they want a very thick solder joint, and yet, if you do the modeling and the reliability testing, you’ll find that even a 20% solder joint in a thick board has more than enough reliability. I’m not advocating that we go down to 20%, but I’m saying let’s get real about what’s actually required. I think we need to move out of the middle ages of soldering in terms of through-hole and consider doing more work in terms of modeling and reliability of the through-hole joint specification.”
“On a through-hole board, that’s certainly the most common problem that we have in terms of soldering,” continues Freedman. “The barrel fill is always an issue. I dealt with very thick boards for a very long time; server boards and telecom boards that are well in excess of 0.093”. They’re just a bear to repair in all regards. The first thing that is spotted generally for repair is the through-hole solder joint itself. We see nice barrel fill on the signals, of course, and marginal solder joints on powers and grounds. Those are always seeing touches. We have operators calling that out because of either visual or X-ray results, and trying to fill things that are already more than strong enough to last the lifetime of the product.”
Ripoli says they quite often find that people who are trying to remove things from the board—especially bottom termination components—and don’t have the proper equipment to do that, tend to rip up pads on the boards. “It ends up in our hands to do pad repairs as well as placing that BGA back down. We have people with over 20 years of experience dealing with those types of repairs,” she notes.
The majority of BEST’s rework and repair involves bottom termination components, says Patten.
“More so than the through-hole,” adds Ripoli. “Definitely, the QFNs, LGAs, BGAs are a large percentage of the work we do. We see a lot of those coming in with excessive solder voiding or solder shorts.”
Price agrees that bottom termination devices are the most prevalent rework requirement. “Often, what drives larger quantities of component rework, is when there’s some issue with a component, whether the manufacturer has determined there’s a bad date code on a component that is populated on hundreds or thousands of boards that needs to be reworked, or there are other issues such as mixed chemistry. An example would be a RoHS lead-free component placed into a leaded soldering process, therefore it doesn’t get to proper reflow temperatures. There are a variety of things that drive large quantity rework applications like that. By and large, BGA, LGA and QFN rework is very challenging to anybody. The rework itself is difficult, as is the inspection process. There’s not a lot of trained eyes that can examine and understand and look at an X-ray with certainty when it comes to ‘head in pillow,’ or ‘solder open’ and other defects, it is very challenging.
“I would say that is probably the most challenging rework application out there for the general population of contract manufacturers that are doing rework in-house, especially when you get into the board designs mentioned earlier with increased thickness of boards and tighter spacing of adjacent components. Preventing collateral issues when you’re doing a hot air rework is difficult, the hot air doesn’t want to stay where it is directed, it wants to spread out, and it does. It will spread out through the board and it will spread out to adjacent components and it will create conditions outside of the target zone that is called collateral damage. Overheating adjacent component, reflowing solder joints and other issues can occur. Add the conformal coating element to it and under fill at that particular rework site, well, now you’ve elevated the degree of difficulty to the rework application. It is extremely difficult to have 100% success rate with all of those factors.”
When it comes to solder voiding, Ripoli says there are many things that can contribute to that issue: one being the customer using mixed alloys when assembling a board. “We’ll find evidence of that going on. It could be that the paste is not the right thickness. It could be the profile on the original assembly that’s causing the issue. To fix the problem, it’s back to the customers to figure out what step in their process is causing an issue,” she says.
“They’re more likely to have the data or the tribal knowledge and the cause of the problem and we might just be able to have the equivalent expertise to solve the problem,” says Patten. “We don’t always get involved in a recommendation. Sometimes, we do and we certainly can, but to learn that entire process for that build and the profiles and so forth doesn’t always go through our hands.”
“We have some customers that work with us while they’re doing their proto stage of assembly,” says Ripoli. “They’ll come back to us and send us a board with connectors and have us measure the solder fill percentage, the X-ray. We’ll give them that feedback, they’ll go back and make changes to their process, send us another batch of boards, and we go through that same process again until they can tweak and find the right solutions to getting that proper fill.”
One of the biggest issues when it comes to rework and repair is the temperature cycles.
“The problem is when you have to rework something, it has to go through four or six cycles of heat or something like that. Then it’s a problem and they didn’t predict that. Nobody predicts reworks on their boards,” says Price.
“Not only that, but you won’t necessarily know that the board has been repaired multiple times before it gets to you,” says Patten.
Most of the time, the rework specialists will never find out how many times a customer—the contract manufacturer/PCB assembler—touched the product before they get it. “It is a big mystery,” says Patten, “and it’s often a cause of unpredictable failures.”
But some PCB assemblers/contract manufacturers do track the number of rework/repair cycle that a board goes through.
“When we get larger quantities of rework or repair, we don’t see that as often; but when we have the one really special custom board, we get that a lot. They tried and they failed, so they give it to us. We don’t know what they’ve tried or how they failed. We don’t get that history usually,” says Patten.
To read the full version of this article, which appeared in the September 2017 issue of SMT Magazine, click here.