While assembling SMT PCBs for a customer, the line unexpectedly ran out of 01005 package size resistors during the production run. This was due to an error on the customer’s part; they simply did not provide us with enough. But the customer still expected the finished product in-hand by the due date, and we did not want to disappoint him.
This, of course, created a dilemma. Should we halt production while additional parts were ordered, or simply continue? We placed a replenishment order immediately, and learned that we could not get them until the next day. We asked, “Is it possible to assemble these boards now, and add the part later when they arrive?” We also had to quickly assess the feasibility adding the part later on. How accessible is the site for this tiny part? Can it be done by a skilled soldering technician? We decided that it could.
An 01005 parts shortage didn’t stop production.
We proceeded to build the order with the exception of that single 01005 component. The site was accessible, we decided, so that the part could indeed be added individually to each PCB. Sure, it would take some time and hand soldering, but it would not cost us nearly as much time as if we had put the entire build on hold to wait for the parts.
Once the needed parts arrived, they were hand soldered onto each assembly by a skilled operator using slender soldering iron tips.
When the parts arrived the next day, we were able to employ our best hand soldering people with soldering irons equipped with special slender tips. The job went quickly and easily, and we were able to minimize downtime on the SMT line due to the unexpected component shortage, and keep this job on schedule, much to the relief of our customer. We came through for them despite their mistake, and they appreciated that.
The 01005 parts were easily and quickly added, and the entire job shipped by the required due date since the rest of the assembly had already been finished.
Efficient use of time, getting whatever can be done while minimizing down time, pays dividends when one is living by tight delivery schedules and expected ship dates. The more that can be achieved without unnecessary delays, the better the chances that a ship date can be honored.
A mirror can bring bad luck, it is said. In this PCB assembly challenge, it certainly did when a mirrored pad layout for a transformer made it impossible to mount the component to its intended location on the top side of a PCB in its usual orientation.
Design error: A mirrored pad layout creates orientation problems between pads and component pins; layout is for bottom-side rather than top-side mounting.
The component’s footprint, it turns out, would work fine if it were on the opposite side of the PCB, but that bottom-side installation is not possible.
Flipped upside down, the SMT transformer’s pins line up fine, except that they are facing upwards. But we can still mount the component and make a robust connection using adhesive and connecting wires.
The customer made a design mistake; although the pads for top-side SMT mounting of the component are in place, they are in mirror-image orientation; e.g., the pad layout with Pin 1 is intended to be installed from the bottom of the board. Consequently, it doesn’t match up in terms of orientation on the top side of the PCB unless the component is literally placed onto its back. But that means that the leads are sticking up into the air, pointing in the wrong direction.
It’s well known that a dab of epoxy can cure a host of ills, and in this case it was simply a matter of dispensing a tiny amount of epoxy onto the back of the component body, in the center, as well as onto its intended location on the SMT PCB assembly.
Small dots of epoxy are applied to the PCB surface and to the component body, before it is attached, the epoxy cured, and the transformer connected pin by pin.
The component is then carefully located in place upside-down and the epoxy cured. With the component robustly mounted in this manner, small wires were then run from each lead (pin) to its corresponding pad on the board’s surface.
It requires skillful hand soldering once the component is in place, but the connection is robust and complete.
In the first half of this column, we began a discussion of the pros and cons regarding the use of conductive inks versus nonconductive inks to fill vias. The images below show cross-sections of a via-in-pad with nonconductive ink on the left and VIP with conductive ink on the right.
Via-in-pad (VIP) filled with nonconductive ink.
Via-in-pad (VIP) filled with conductive ink.
In that column, we discussed a design that required a nonconductive ink in the through-hole via and conductive ink in the blind via. Now we ask, what were the drivers behind this decision? Why would one use two different types of inks in vias in the same PCB, and why conductive vs. nonconductive ink? The answers are actually a bit more complex.
Copper plating is one factor, as an example. For years it has been generally accepted that copper plating is not a viable substitute for ink (conductive or nonconductive) to fill a through-hole via, buried via, or blind via. It was believed that to plate a via shut and to cover the surface with copper would take “forever,” relatively speaking, if it could be done at all.
One reasoned that not only would the process of plating the via shut with copper be prohibitively time-consuming, but even if it were technically possible to fill the hole with plated copper, the unwanted consequence of plating so much copper in the hole would result in excessive “button” or surface copper height that would lead to other defects and/or reliability risks.
Blind via plated shut with copper.
Nowadays there are several efficient processes for copper plating to fill vias; two of these are pulse and DC rectification. Some require button plating, a two-stage process; others have evolved to the point where a single-stage panel plate will fill certain via structures while depositing less material on the surface, thereby leaving a manageable surface copper thickness. In this way, one can continue to produce a high-density product without the need for a secondary ink-filling operation.
Further, there are solutions to filling micro, blind and buried vias that require no additional process time or steps; e.g., resin or B-stage fill. The consensus was that it wasn’t possible to do this reliably. While conventional prepreg (B-Stage) historically struggled to fully and consistently fill vias, there are now specialized prepregs and bonding materials specifically engineered to do just this process reliably.
One laminate company produces a series of FR-4, lead-free, polyimide, low-loss and other high-performance laminates and prepregs. Within their product line they offer a sub-set of prepreg (B-Stage) called the VF-series (whereby VF is an abbreviation for via fill).
Via Fill (VF) prepreg product, where core and prepreg are combined to create a pure, homogeneous material package.
Where we have the instance of a buried via filled with one ply of VF material, the blind via is fully filled with resin, and the dielectric distance between outer foil and inner sub assembly is very uniform. In this case, the VF matches the family of core and prepreg it is combined with, so that it permits the creation of a pure homogeneous material package, eliminating the need for a hybrid material / laminate package. The VF prepreg has been engineered for enhanced rheology and filler content so that during the lamination process the blind and buried vias found in a sequential lamination sub assembly will be fully filled.
VF Prepreg is just one example of available materials designed to fill vias during the lamination process, thus eliminating the need for a secondary operation. What process and material should you use? To make the best decision, you need to understand not just what result you want to achieve, but why.
Not long ago I had an application involving a customer’s requirement of a specific brand of conductive ink to fill a small through-hole via. The assembly was a double-sided PCB on a relatively thin (0.010″ thick) PTFE/Teflon material.
The ink-filling process requires a planarization or sanding operation after the ink is cured in order to remove excess ink from the copper surface. The planarization process always includes some inherent risks and/or limitations such as:
Dimensional distortion of the panel of PCB material.
Imprecision, resulting in uneven copper thickness and poor control of circuit etching.
Reduced peel strength of the surface copper.
In this case, all these negative aspects of planarization were amplified because the material was a soft; thin Teflon with RA copper. This material is highly unstable to begin with and susceptible to distortion.
The PCB manufacturer struggled to meet the customer’s requirements, but excess cost, time to produce, delays, and lower yields resulting when compliant product was finally produced were a real problem, prompting further discussion with the customer.
A breakthrough occurred when we began to ask why we were using certain materials and questioned the necessity and benefit of each step in the process. We realized that the via filling; i.e., the specific material requested by the customer, was being used to prevent solder from flowing through the vias during assembly. But what else was it there for?
After critical examination, we realized:
That there was no need for conductivity in the filling material , let alone any reason for it to be limited to the customer’s specifically preferred ink material.
There was no need for a copper pad to be plated over the surface of the material or via, since nothing was being soldered on top of the via.
There was no need for a specific brand of ink material.
Alternative materials and processes could therefore be explored.
After all, we began to examine the real purpose that the via filling was intended to address, and more importantly, what it was not there for. The material had been used, all along based upon a group of assumptions that, when examined, weren’t true and did not justify the use of that specific (and costly) ink material. Its use simply could not stand up to challenging questions, such as added reliability, electrical advantages or mechanical aspects or even thermal characteristics or properties. It contributed to none of these justifying criteria.
Buried via fully filled with resin; note that the dielectric thickness between the outer foil and the inner subassembly is very uniform.
In summary, when evaluating a new product, manufacturing process, etc.:
Challenge any long-held assumptions.
Gather information from multiple sources.
Qualify that what you have been told by others is really best for your needs and not skewed merely to support the choice of a specific product.
Manufacturers must talk with the designer to understand what designers really want to accomplish and why. Designers must speak with manufacturers in order to understand the intricacies of the process. Finally, as technology evolves and more innovative solutions for emerging applications or enhanced solutions for existing ones are found, cooperation and collaboration are the keys to optimizing decisions and selections.
This PCB assembly challenge involved attaching a solar panel to one side of a pad using solder paste with a pass through an SMT reflow soldering oven.
Solder wicking through the unmasked vias to the back side forms unacceptable “bumps” on top of the vias.
The attachment or bond itself wasn’t the issue; but after the first trial runs, it was clear that solder wicking through the unmasked vias was going to be. Solder would wick through the unmasked vias to the back side and form “bumps” on top of the vias.
These bumps made the surface nonplanar and of course were unacceptable. It wasn’t an issue of using excess solder paste. But the “wicked wicking” had to be stopped, or at least prevented.
Kapton tape is applied to cover the unmasked vias; it will block the molten solder from leaking through.
But how? Clearly, to keep the solder where we wanted it to remain during reflow, we had to find a way to prevent it from wicking up, collecting at the opposite ends of the vias and forming bumps. We had to find a solution that was simple, temporary, and tolerant of reflow soldering temperatures. The answer was Kapton polyimide tape, a familiar product to PCB assemblers for many years, and a material that does not degrade at reflow temperatures.
Kapton tape was applied to cover the unmasked vias in order to block the molten solder from leaking through the vias to the back side during reflow. After reflow and cooling, it was a simple matter to peel off the tape. This temporary masking solution worked; there were no more solder bumps on the back side of the assembly, and the cost of the fix in terms of time and material was very low.
Figure 3. This temporary masking solution worked; there are no more solder bumps on the back side of the assembly.
This is a little bit like the old college prank of trying to see how many kids can squeeze into a telephone booth. Pretty soon everyone’s too close for comfort!
In this PCB assembly challenge, someone made a mistake and created a layout for rows of dual-flat no-leads (DFN) SMT packages without taking into account the size of the component bodies. The footprints are too close together, and the bodies of the components are touching.
Because they don’t all fit, as the packages are lined up there isn’t enough room, and alignment issues develop for some of the IC locations. They’re forced off their footprints, while others appear to be acceptable.
Figure 1. With DFN footprints too close to one another, component bodies are actually touching and causing alignment issues, literally forcing others off their footprints.
As can be seen from the photos (Figures 1 and 2), the crowding causes alignment issues for locations IC1, IC5, IC7, IC9, IC13, and IC15. Locations IC3 and IC11 seem fine.
What can be done? It’s too late to redesign and order new PCBs, and there is no possibility of shrinking the dimensions of the components.
Figure 3. Removal of components in locations IC5 and IC1 have allowed the rest to fit properly.
Luckily, the customer had a solution that worked: removal of the components in locations IC5 and IC1 (Figure 3). This permitted the remaining parts to fit correctly; it made “breathing room” for the rest, and best of all, was accomplished without compromising the functionality of the circuit.
If the shoe doesn’t fit, can you still wear it? You might have to if they are the only shoes available. In this case, the SMD packages for this PCB assembly application are actually wider than the PCB footprint itself. There are any number of reasons for this, from changes in component design to substitution issues, but we won’t get into that here. But the problem is that the leads actually overhang the SMT pads and extend onto the solder mask area (Figure 1).
These packages are too large for the corresponding footprints, with leads extending and overlapping the solder mask.
This, of course, is unacceptable. But attempting to shorten or “snip” the leads won’t work either; the shear force could easily be too much for the package’s integrity.
The solution was to bend the pins in slightly so that they could fit onto the SMT pad without extending or overhanging off of the pads (Figure 2).
Bending the leads back slightly to fit within the confines of the pads is the only acceptable solution.
Certainly some stress and tension is applied in mechanically bending the leads, but not enough that we need to worry about it. And even though the lead is contacting the pad at a changed angle, there’s enough solder to create a robust solder joint. Remember that in the early days of SMT, some through-hole DIPs were snipped off and soldered to SMT pads creating butt joints, and these proved to be robust and reliable.
The bent leads solder to the pads just fine, forming robust solder joints, and meeting acceptability criteria.
An added advantage of not shortening the leads is that retaining lead length provides added spring-like flexibility for the lead to flex with thermal cycling, minimizing the possibility of solder joint failure due to thermally-induced stress. It isn’t much trouble, a good solder joint is created, and the part passes standard acceptability criteria because, in part, the leads are contained within the solderable pad area.
Figure 1. Two SMDs have misaligned during reflow due to uneven pad sizes and disproportionate solder liquidus surface tension.
Figure 2. Temporary solder mask glue applied to two corners of each component prior to reflow keeps them in place.
Figure 3. Post-reflow, the adhesive is easily removed, and the SMDs are perfectly positioned as they should.
The surface tension of liquidus solder exerts a considerable pulling force on a component during reflow. This is why, once upon a time, small components could be relied on to self-align on SMT pads during reflow. They still can, of course, providing that all things are equal, such as pad dimensions all around. But if they’re not, you can expect problems.
In this case, two components had shifted away from their center location on the PCB footprint during reflow (Figure 1). This was due to the fact that a large SMT pad on one side of the components, see photo, was exerting a stronger pulling force on the component than the ordinary-size pads on the opposite side. More surface area means more pulling force, and consequently component misalignment. It doesn’t matter that the placement machine put the part in the right location beforehand.
Mechanical fixturing simply wasn’t a practical solution. Instead, two dots of temporary solder mask glue were applied to the corners on one side of each component, prior to reflow, to hold it in place. The glue acts as a temporary adhesive and prevents the parts from moving during reflow because it is stronger than the pull of the liquidus solder. After reflow, the glue is easily removed, and the SMDs are perfectly centered. Problem solved!
Figure 1. Original part’s through-hole pins are not long enough to go through the board and be soldered from the bottom side.
Figure 2. Cross-section of PCB showing penetration depth of part’s original pins.
Figure 3. Side view showing new pin length (inset: bottom side view).
In another PCB assembly challenge, the customer’s BoM called for a through-hole header part to be installed on a circuit board, a mixed technology (SMT and PTH) assembly. A problem became immediately apparent with the first PCB that our operators began to assemble; this was a very thick circuit board (12.15mm thick), but the part specified and received did not have terminal pins long enough to protrude all the way through the PCB to other side (bottom side) (Figure 1). Obviously, we needed the pins to penetrate all the way to the bottom side and protrude so that they could be properly soldered using wave or selective soldering techniques.
What to do?
In this instance, a replacement part with pins of sufficient length simply wasn’t available. The easiest solution (although labor-intensive) required manual removal of all the original pins from the part, and their replacement with longer pins of sufficient length, actually, to form good and robust solder joints on the bottom side of the PCB assembly (Figure 3). Once the longer pins were added, it became a simple matter to re-insert the through-hole header part and solder it in place from the bottom. The parts of the pin protruding from the solder joint could then be dealt with the same way as any other soldered through-hole pins.
Figure 1 shows a closeup photo of a PCB assembly, it seems as though solder has flowed “down the drain” and away from the solder joint where it’s needed.
In fact it has, because the customer has inconveniently located a via right through the center of one of the two topside SMT pads for a surface mount component. When the assembled PCB is run through reflow, the molten solder drains away through the barrel of the via and out the other side of the PCB. There isn’t enough solder remaining post-reflow to create an acceptable solder joint per IPC-A-610. The joint is “starved”; this is unacceptable. What to do?
Figure 1. Insufficient solder, i.e., “starved” solder joint on an SMD pad.
The via is there to stay, by virtue of the customer design. So, no matter how many times solder is added to the joint, every time the PCB is run through the reflow oven the solder is going to drain away because the PCB , including the via, is at reflow temperature.
Obviously, more than one run through the oven makes no sense. The only practical solution is to manually add solder to the individual solder joint, post-reflow, without running the entire PCB through another thermal cycle. It’s a touchup procedure that’s required to create a robust SMT solder joint that meets acceptability criteria. This is a manual PCB assembly soldering process that should be performed by a skilled hand-soldering or rework operator. Solder is added only to the joint, via cored wire solder or solid wire with flux, in order to build up the volume of solder at the solder joint to provide strength, connectivity, and an acceptable meniscus per IPC standards, covering the via drain-hole. The solder won’t flow through the via because only the surface joint area is heated.
Figure 2. The solution: Add solder to the joint manually via a touchup procedure.
It may seem tedious, but a skilled operator can touch up the joint in a few seconds, and if there is only one instance per assembly it won’t appreciably cause production delays.
Sometimes, in PCB assembly, it’s not the layout of the SMT PCB that creates issues, but the design of the part itself, or the plan for the part’s location, given its dimensions. We have to ask ourselves, sometimes, “What were they thinking?”
In this case, a customer’s BoM called for a part (an RF200 module with through-hole pins) to be mounted onto a PCB. At one end is a bulky SMA connector that due to its size exceeds the thickness of the module. The SMA connector faces inward on the PCB; it’s not mounted to hang over the edge. As a result, the SMA connector bottom side touches the board and props one end of it up; it doesn’t permit the module’s pins to be properly inserted into their corresponding PTH barrels on the board. One end of the SMA is pointing upward on an angle like a missile-launcher.
This is obviously not acceptable in circuit board assembly; not only is customer access to the connector compromised, but the module cannot be mounted in a planar fashion and having some of the pins fully inserted and some halfway out of the barrels, with one end of the module elevated, is certainly not acceptable.
The fix was relatively easy; we recommended that the customer allow us to use two single-row socket pin adaptors to provide the standoff necessary to keep the SMA connector from touching the board while at the same time allowing easy and unobstructed access to the connector.
Two socket headers were used, corresponding to the module’s two rows of pins. Not only did this provide the needed standoff, without creating any other issues, but it also allowed the customer the potential for removing or replacing the module in the socket pin adaptor in the future without serious rework issues, since it’s a mechanical mounting. It’s also a robust electronic assembly connection in terms of strength and durability, and the module is completely planar with the surrounding PCB surface.