Stencils: Not As Simple As They Seem
Stephanie Hardin of Integrated Ideas and Technologies Inc. discusses her role in the supply chain as a stencil manufacturer, improvements she sees from micromachined step stencils, and why she believes trying to have standardized stencil layouts is wishful thinking due to the many fluctuating variables.
Nolan Johnson: Stephanie, can you start by introducing yourself?
Stephanie Hardin: I have been with IIT, a manufacturer of stencils, for 19 years. I started as a stencil design technician, and I’m currently vice president and part-owner. From the stencil design standpoint, I’ve seen a lot of changes in components, and they’ve done nothing but shrink the entire time. We utilize an industry-leading fiber laser cutting system that has a complete operator lockout, which is important because the operator can’t make any adjustments to the beam or focal points. The ability of less-equipped machines to adjust those variables creates the potential to produce a less than perfect cut. With small devices, a clean aperture wall that meets the ±0.00015” tolerance requirement is essential.
I work with clients on a regular basis with high-density, high-mix, low-volume builds where there are micro components in the same area as components that require a much more robust solder joint. It’s always about balancing the aperture design and the stencil thickness to print the volume required to create a good solder joint for all devices.
Johnson: It appears there is a tendency to put insufficient solder paste down in these smaller spaces, and then the joints don’t work. There’s also the issue of having to use a thicker stencil, which means more possibility to have issues around the edge of the stencil cutout. How do you balance those things?
Hardin: If you use a thicker stencil and get the paste to release, then you’re also potentially dealing with the issues that arise due to excess solder volume, such as mid-chip solder balls, tombstoning, and bridging, to name a few. Primarily, when we look at these stencil designs, we design with the smallest device on the board as the focus. We then expand the criteria to create a design and stencil thickness combination that’s going to work for everything else on the board.
Before the introduction of step stencils, inevitably, you were doing touch up somewhere. You were making your stencil thick enough to deal with the majority of the components and hand placing the smaller things, or you were doing the opposite and having to touch up the joints on the bigger stuff. It required a lot of extra handling to get it done, but with the step stencils, you can achieve all of it in a single stencil. You are now able to have recessed areas in a three-mil thickness to print a 02001 component, and then you can have a six mil build up on the same stencil for a large connector.
Johnson: I’m not that familiar with a step stencil. How does that get put together?
Hardin: There are a couple of ways to manufacture them. They used to be chemically etched or electroformed, both of which were rather expensive options. Laser-welded stencils are a newer technology, where they take thicker pieces of material, build it up, and weld it to a thinner piece. The latest technology is micro-machining. The micromachining process leaves a much smoother transition in the steps where the laser-welded stencils left a hard 90° angle. Imagine your squeegee trying to move across that surface, and when it hits that 90° angle, it might as well hit a brick wall. It would cause the squeegee to skip up over and potentially cause misprint in that step-up area, at least for part of it.
With the micromachined smooth transitions, the squeegee moves more freely over the stepped area. Two things happen there: less squeegee damage, so your squeegee lasts longer, and the stencil also lasts longer. You don’t have multiple layers that can delaminate in production.
Johnson: It was recently pointed out by somebody working on stencil technology that the features are becoming fine enough that they're even starting to worry about the metal grain and its effect on the application. Is this is something that you find yourselves working with regularly to that almost-microscopic level of detail?
Hardin: Definitely. Some of the aperture openings that we’re dealing with are eight-thousandths of an inch. As far as looking at the physical component goes, it would look like a fleck of pepper on your finger if the component happened to be black in color. That’s the only way you’re going to see it, if at all. This presents issues with paste release. The fine grain stencil materials available now improve paste release before any other coatings are applied, such as a nanocoating. Nanocoatings have also assisted a great deal with enhancing paste release. Nanocoatings also allow for more prints to be completed before the stencil needs to be cleaned.
Johnson: Are you getting stencil patterns delivered to you by the customer that are workable right off the bat, or do you have to do a lot of stencil pattern prep?
Hardin: Typically, there’s quite a bit of prep, unless it’s an OEM that has a set product and has had the time to go through and develop their process and their design library, to achieve the best results in production. Often, OEMs supply their data already modified by in-house designers, that are able to utilize a set design library. Their process is controlled each time; it’s their product, and they build it. They are not dealing with the variations created by customer-specified solder paste, three different types of 0402 capacitors, or a variation in board finish that is inherent when obtaining boards from multiple sources.
The other side of that is the contract manufacturers who, oftentimes, see these new components for the first time and ask us, “What do we do here? How are we going to print this?” They look for recommendations from us on the best design to net a first-time pass. Their client may have supplied the boards, the parts, and specified a stencil thickness. The contract manufacturer may also not have the authority to make any changes to the stencil design, stencil thickness, or solder paste that the client calls out. The complications can compound quickly.
Oftentimes, the footprint that is generated from a generic design library might not work in all instances. The often-overlooked factor when considering aperture design is the actual production environment. When devices are this small, any change in humidity, printer setup, or solder paste management could throw a standardized aperture design layout out the window. You have to be able to design for who’s building it and how their line is set up.
Johnson: In other words, following the guidelines in the datasheet for setting up your pads and your stencil patterns is risky.
Hardin: It gets you in the neighborhood. It usually comes down to starting with the manufacturer-recommended layout for the first build. Then, if issues arise during the build, get back with your stencil manufacturer on what problems you were having and start to modify the design to alleviate the issues. Each manufacturer will recommend a stencil thickness for their device. The trick is taking into consideration the recommended thickness on all devices and creating a stencil that will work.
Johnson: From your perspective, how possible is it to look at the data and know that there's a problem with the stencil upfront? Can you be proactive about modifying that stencil file, or do you have to wait for a person to first article to figure that out?
Hardin: I can be proactive about it to a point. I can look at the design versus the stencil thickness that the client’s asking for. If a 5-mil stencil thickness is requested, I can identify the components that might have problems with that thickness. Even if the device is new to me, I’ll be able to tell if there’s not enough space between the lands, which might create a solder ball issue under the device or not enough of a gap between leads on a multi-lead device that may create a bridging issue. I will also be able to identify concerns with ground pads under leadless devices and recommend a design to mitigate part movement during reflow.
Johnson: Since designers can't trust the datasheet, what else do they need to do?
Hardin: That’s a tough question because there are so many other variables in the production line that contribute to the success or failure of creating that joint. If there’s a different paste being used? Or is the paste is sitting out too long, and the flux is evaporating out? Is there sufficient under board support with the proper combination of squeegee pressure and speed? Are you printing on contact or off contact? It would be difficult for the designer to say it should look like this under optimal conditions where the condition is different in each facility. I’m not sure how they would do that.
Johnson: It sounds like this is highly dependent on the situation.
Hardin: From my experience, based on client feedback with printing issues, it’s difficult to pin that down. On some of the older multi-lead devices, it was a pretty easy thing to look at the specification and say, “They’re calling out a pad that’s awfully long. I’m going to have either a bridge between leads or solder balling out on the toe or heel of my device if I print this entire pad the way it’s designed.” You would have to shorten up the pad a little bit. It is much more difficult to determine ahead of time for the micro-devices. There’s not as much leeway in the manufacturing process to compensate for any other issue because, along with the device size, your process window keeps shrinking and shrinking.
Again, if you were talking about an OEM that had a 100% controlled process and environment where nothing ever changed, it would be easy to work with that OEM and ask, “Where does this pad layout need to be to be successful?” Make the change and go on. But when you get into contract manufacturing, it’s a completely different environment.
Johnson: As a stencil supplier, imagine you’re working with a customer, and all of a sudden, something changes in their process. Things aren't working. They need to make modifications to the stencils. Are we talking about relocating cutouts in the X, Y axes, and changing to different stencil materials or sizes? What are some of the ways you can handle some of those problems?
Hardin: All the things you mentioned. I’d look for specific defect information from them. And from that point, it could be shrinking the pad or changing the pad geometry into a home plate or a reverse home plate. It could also be changing the stencil thickness in that location—anything to affect the volume to net the desired result.
Johnson: Is it commonplace to have this happen at a production customer?
Hardin: It can be initially. Stencils houses do quite a bit of design work for clients to try to get around these things. Again, when you’re dealing with contract manufacturing, it could be that the client’s going to build that board once, and they might not ever see it again. It could be that they’re building that board once, and then there are plans for big production, so they’ll spend the time to go back through and work on any defect data that they had to improve their yield the next time.
Johnson: Do you design work specific to the design of the stencil, or are you going back into the board design itself and suggest changes?
Hardin: Usually, making changes to the stencil is the quickest, easiest, least expensive way to affect a positive change in production. Sometimes, in extreme cases, it comes back to having to go get the boards revised. There could be factors that a stencil can’t really address, such as masking or silkscreen legend that’s holding the stencil off contact.
Johnson: We’ve been learning that legend and solder mask planarity are becoming precise enough that even that matters now.
Hardin: That’s why it’s a tough question to answer. My instinct to say it would be impossible for them to figure that out upfront, but impossible is not a word we like in our industry. However, it is extremely difficult for sure. To have the expectation that a board designer is going to be able to have a single component layout in their library, the stencil cut one to one with the data, and everything’s going to go great is simply not realistic.
Barry Matties: How often do you have to rework stencils? Is this a pretty common occurrence, or is it a rare occurrence?
Hardin: Bringing already-cut stencils back in-house for rework is pretty rare. If the customer goes into production after the first article goes through, and had some issues they want to address to improve yields, then we’d go through specific changes, but that’s not as common. We try and spend that time upfront, in the initial design, to mitigate rework for everyone.
Matties: Why don’t more contract manufacturers or assembly houses have their own stencil cutting capabilities?
Hardin: That’s a good question too. I think the cost of the equipment sitting on the floor for the amount of time that they’d be utilizing it would be a big negative to doing that. And then you’d have to have someone in the facility with all of that design knowledge on top of everything else you have going on to do that. Primarily, it would be having a piece of equipment sitting on the floor underutilized. With Lean manufacturing being our focus, floor space is valuable. That would be a big dead space right in the middle of your floor that did not generate regular revenue.
Matties: It seems like you're putting a lot of your experience into making the stencil right the first time, and your only focus is the stencils. I'm surprised that more and more people don't have it. What you're saying makes sense; there has to be a crossover point where they're producing so many stencils that it makes sense for them to have it in-house. And yet some of those people don't.
Hardin: If you take into consideration your average stencil house is putting out somewhere in the neighborhood of a hundred stencils a day, to fully utilize that equipment and how fast it is, it’s going to be a rare manufacturer that needs a hundred stencils a day.
Happy Holden: Pad definitions must be a challenge.
Hardin: Yes. It’s inconsistent. You would be surprised at how many different pad layouts there are for an 0402 capacitor. Different part manufacturers and board manufacturers are never the same. Even if it says, “I want to have a standardized layout for an 0402 aperture,” it’s only going to work if every board you get in has the same kind of layout for that 0402 aperture, and that never happens.
Holden: How quick can you build a stencil if everything goes right and how long does it take if everything doesn't go right?
Hardin: If everything in the data looks good, and granted I’ve been doing this a pretty long time, I can get through a simple design in five minutes and have it in production. If it gets a lot more involved, with a lot more custom designing required, it could take me an hour to do the design work on it.
Holden: Does the turnaround time to the customer vary that much?
Hardin: I have a job now that’s going to take me probably two hours to do because of some custom fencing that they want to place solder paste on the board. There are a lot of things to get around, components up near the edge of the fence and through-holes that they want to avoid. Things like that can take more time to draw out. But in general, the stencil will still turn the same day even if we’re making quite a few changes to it.
Holden: That’s pretty good. The problem would become if one after another, they were all a headache; then, it would start to back up.
Hardin: That would be a little more of a problem.
Johnson: There has been a lot of work in solder paste lately, including a number of new products delivering new capabilities. How much work is involved for you in figuring out what to do with the stencil to accommodate a particular kind of paste?
Hardin: That doesn’t factor in as much to what we’re doing, believe it or not. There was a huge change made when people were switching from Type 3 to Type 4 paste because the spheres were smaller, and the volume was a little bit more volume. That and the switch from leaded to lead-free paste have been the most significant changes that I’ve had to design around.
For the most part, the paste type doesn’t get relayed to me unless to specify that it’s leaded or lead-free paste, and that’s because lead-free paste is not as forgiving in its travel. It doesn’t like to move. Where you put it is where it stays, and you can’t get too out there with your design. It also has a tendency to solder ball when it’s printed on the mask, so you have to be pretty careful about that too when you’re looking at moving pad geometries around.
Matties: You must be able to look at a stencil and read it as if it was a printed language.
Hardin: It’s getting there.
Matties: When it comes to defects, do they ever come back to the stencil and try to assess blame on the stencil itself? Is there any liability risk for you and your business, or is it not that level?
Hardin: We will accept some responsibility for that if we’ve gone through the design and our customer’s given us their confidence that we’re evaluating potential errors for them. If we miss narrowing up the width on a fine-pitch QFN device, and that wider aperture in the stencil results on a bridge or part floating, we will make the change and replace the stencil at no charge. That’s one of our services.
Matties: I'm thinking down to field failure, and the way traceability is going in liability is shifting. We're such a litigious society these days.
Hardin: I’ve only heard of one field failure coming back to a stencil design issue, and that was with respect to the volume of solder on a ground pad underneath a lead-less device. When those lead-less devices first came out, manufacturers were recommending about a 50% coverage, and now that they’ve been out in the marketplace, 70% coverage is what’s needed to keep those functioning. It was quite a long time ago now, but there wasn’t any product liability against the stencil manufacturer.
Matties: It sounds like yield and reliability rest on the stencil. For the stencil, manufacturers are relying on their external partners to make sure that a key step is correct and accurate.
Hardin: A lot falls into that part of the process.
Holden: Is there ever a situation in which you have to get back to them and say, "We can't figure out how to do this,” or, "You're going to have to do this with two stencils in two passes because you know what's required for this board covers such an extreme, and we don't know how to do it with one stencil or one pass."
Hardin: The only time I can think of that happening was when I had an aerospace board that came through that had 12-ounce copper buildup in places, but not across the whole thing. There were 0402 components backed right up to the edge of these big plated up areas. But with the micromachined step stencils now, you can have steps on both sides of the stencils.
Holden: One of the things that designers are slowly learning is when they're designing a multilayer, they must get in touch with their fabricator to ask about the material, and the fabricator will say, "Send us your stack-up and everything and something about your circuits so that we can determine what the as pressed thickness will be." Otherwise, you can't determine your line width for transmission line-controlled impedance. Unless this is an absolute duplicate of the way we did it before, you'll end up with the wrong impedance. It's not the fabricator's fault; your drawing is wrong, so do you have a set of checklist or pre-checks of or do’s and don'ts of best practices that if they can follow those guidelines or contact you early, it's going to be much smoother sailing and you're going to have a nice stencil that turns around quickly. That's not going to be a source of your problem with assembly.
Hardin: Especially with new clients or someone that’s new to stenciling altogether, often, they say, “I don’t know what to do with this.” I say, “Send me your file. I’ll go through it and make recommendations on what’s going to ease your manufacturing concerns.” That’s the route we usually take.
Johnson: Yet, on the surface, it seems like the stencil would be simple.
Hardin: They used to be a long time ago.
Johnson: Is there anything else that you’d like to address?
Hardin: We talked about that misconception that you could have a standardized design library that’s going to solve everybody’s problems. That is a misconception at best. Even if you had every designer and board house turning out the exact same layout for certain pad or device, we’re still going to have all these environmental factors in the manufacturing facility itself. It’s not going to be cut and dry.
Holden: You and the assembly people are trying to solve the problem without having to go back to the designer.
Hardin: That’s right. By the time they order this stencil, they have a production schedule to meet. That stencil has to go on the line in a day or maybe two. They’re getting their kits together and need to build and ship, so there’s often not time in that schedule to go back to the designer and try to work it out.
Holden: If the designer opens it up because if that phase of the project gets closed, they’re onto the next design.
Hardin: Exactly. On paper, it should work fine, but it doesn’t always work that way.
Matties: Who wants to take the cost of a re-spin as well at that point? As you said, they’re ready to put the parts on it and ship it all. They’re not interested in looking at circuit design. They need you to solve the problem.
Hardin: Right. It’s a quick turn business these days. Get it done, and keep everybody moving and making money.
Holden: Do you see anything in the future that will help you out or improve the situation?
Hardin: Not really. This has been something we’ve been dealing with quite a bit for a number of years, and I know the devices are going to continue to shrink, and we’re going to have to continue to make adjustments for that. We’ve been playing this shrinking game here for a number of years now. I remember when a 0402 was the end of the world, and everyone thought, “How are we going to print this? How are we going to place it?” Now, those devices are common.
Matties: Thank you for speaking with us.
Hardin: Thanks for your time.