Abstract
Surface mount technology was developed in the 1960’s and implemented on printed circuit boards in the 1980’s. High electronic assemblies are dominated nowadays by surface mount devices. The number of throughhole components left on the assembly declines continuously. Thishas a major impact on the assembly process. Many through-hole devices are soldered with pin-in-paste solder technology in a reflow oven, but for some this is no option. The reason for this can be the higher process temperatures of a reflow process or the lack of space for higher solder paste deposits on the board. Liquid soldering of through-hole leads remains a very robust method of making very reliable and strong connections of the components to the board. However, a point-to-point soldering process consumes too much time for a high-volume product and dip soldering comes with additional tooling that makes the production line less flexible. To meet short cycle times and still be flex- ible, a dip technology is introduced on an inline machine. This sustainable method of soldering is possible due to new advanced nozzle designs made possible with new 3D printing technology. Soldering methods are compared using Design of Experiments to determine throughput, as well as energy and material consumption. Several experiments contain methods to eliminate bridges and optimize heat transfer by improving the design of nozzles and solder flow dynamics and reducing cycle times.
Introduction
Most components have a SMD version to enable reflow soldering. For some, through-hole is still used mainly because of the strength of a through-hole connection that is approximately 20 times stronger than SMD alternatives. Through-hole components can also be reflowed using a pin-in-paste process, but the assembly requires enough free space to apply more solder paste. Some applications have staggered stencils to combine fine-pitch with pin-in-paste components. Even preforms are used to get enough solder in the barrels. The major process challenge with through-hole soldering is getting sufficient hole fill (no voids). Some paste tends to drip from the pins in the first heating zones. This not only contaminates the oven but also results in an open
or insufficiently filled barrel. Alternative methods for pin-in-paste soldering are wave and selective. Wave technology is still there in the field but is a less sustainable soldering method. The consumption of nitrogen, solder, and energy are significantly higher than selective soldering. The use of pallets or carriers should also be eliminated because they need to be cleaned, are dedicated for one product, increase production costs dramatically, and require additional handling. One benefit of a pallet is that it keeps the board flat.After one or two reflow cycles printed circuit boards tend to warp which is a concern in the selective soldering process. The second issue with reflowed boards is that the solderability decreases after reflow cycles. Therefore, it is recommended to use an inert reflow process. The use of nitrogen in a reflow oven benefits the wettability in any subsequent soldering process. What is left after reflow soldering are headers, board to board connectors, board in board, elco’s, wires, and special applications. The presence of electrical cars and batteries comes with heavy components and high thermal mass pins that needs to be soldered. This, in combination with fine-pitch components on one assembly, is a challenge for the soldering process. This selective soldering process requires wave formers and nozzles that have the capability to transfer sufficient amounts of energy into the barrels in order to get the solder flowing to the solder destination side.
Selective Soldering Methods
The selective soldering methods that are available today are driven by throughput. In high-volume products, like there are many examples of in automotive industry, the dip process is often
applied. In a dip soldering process the boards are dipped into a large solder pot that has static nozzles on the spots where the through-hole connectors are located. The dip time is 8 seconds
maximum (to avoid material damage) which results in a cycle time of approximately 30 seconds per assembly. The downside of this process is the dedicated tooling required for each assembly.
From a process point of view the challenges in selective soldering are always the same:
• Good hole filling
• No bridging
• No touching or remelting of SMD components
• No solder balls
• No electromigration
The last two are mainly defined by flux and solder mask. The focus for soldering is on hole filling and bridging. The alternative for a selective soldering dip process is the use of a wave
solder machine in combination with pallets. Here all SMD components are covered by the pallet. This method is still used by smaller companies that keep their technology limited to reflow
and wave soldering. Downsides are the high costs for the pallets and handling and cleaning them. Energy and solder consumption are also significantly higher, so this is not a sustainable
method. The problems with hole filling are less because a wave can bring a high amount of energy in the solder joint. Solder temperatures are 260°C for lead-free applications, typically
about 50°C lower than selective soldering. Depending on the thermal mass of the assembly, cycle times can be 20 seconds orequal to d ip soldering. All discussed methods so far need special tooling or pallets. This is very expensive for small batches and low volumes. The products that need to be soldered are so variable that there is no one solution. Point-to-point soldering is the most used. It doesn’t require special tooling and the only variable is the dimensions of the nozzle which may change from one product to the other or there may be multiple nozzles within one product. The mainstream is using wettable nozzles. Most driven by space, but many application engineers are following the mainstream and have no time for experiments due to the production time pressure. The pros and cons are known by selective soldering engineers.
What Application Is Required for Next Generation Products?
There are trends in board assembly and solder materials. New plastics that are more compatible with higher temperatures makes it possible to convert some through-hole components to reflow
soldering. Another trend is more fine-pitch components and more functionality. As an example, in one of the consumer products that is dip soldered in high volumes, a USB connector is replaced by a fine-pitch double high-speed USB. To overcome bridging several modifications have been applied. These include a reduction of pad dimensions, introduction of silk lines, and when necessary for some applications screens are implemented, see Figure 1. Another trend is completely the opposite, instead of smaller components, power electronics and the introduction of electrical
vehicles and autonomous driving come with ultra-capacitors and other heavy metal parts that needs to be soldered. Fine-pitch requires small nozzles that use limited space and heavier compo-
nents need nozzles that can bring in a lot of heat. On the machine side of selective soldering the trend moves to multiple solder pots with easily exchangeable nozzles. If a solder
machine has two or more solder pots the job can be split. One pot for fine-pitch components using a smaller nozzle and anotherpot w ith large nozzles to bring sufficient heat in the board and enable soldering of high thermal mass components. Nozzle technology is making big changes. Metal processing from solid steel is being replaced by 3D printing of the nozzles that can be
printed out of stainless steel or Titanium. Most likely a significant portion of printed circuit boards will be soldered on selective solder machines with multiple solder pots that can drag and dip on dedicated nozzles. They will have larger nozzles in a small solder pot in which headers and heavy metals can be dipped to save cycle time, see Figure
2.
Nozzle Technology and Experiments
Both examples already show the challenges of the selective soldering process. For fine-pitch the risk for bridging is high
and for the heavy components a sufficient hole fill is required. The design of a nozzle is playing a significant role in the hole fill. The goal is to bring enough heat into the solder area to allow the liquid solder to flow into the barrel and all the way up to the solder destination side without solidifying. In soldering the heat-
ing of the circuit board is critical to the soldering quality.The Ishikawa diagram shows what parameters impact the heat
transfer to the solder joint, see Figure 3. The goal of the study was to set up an experiment to determine the amount of energy that the different nozzles transfer to the circuit assembly. The wetting is defined by the solderability of the metals (no oxidation) and the heat required to allow the solder to flow to the solder destination side. The flux and nitrogen are materials that eliminate oxidation. Flux must clean the surface and together with the nitrogen prevent oxidation during soldering. A laser sensor was used to determine he topside of the solder nozzle. From there the pump frequency was increased to a solder wave height 3.0mm above the topside of the nozzle. The same method was used to define the frequency of the electro-magnetic pump for a wave height of 4.0mm. This procedure was repeated for all the tested nozzles. The non-wettable nozzle has the solder flowing in the opposite direction of the soldering. This is identical to a wave soldering process. After defining the 3.0mm wave height the solder that flowed out of the nozzle was collected in a bin. The solder amount was measured on a balance after it was solidified and cooled down. The time was recorded by video the experiment. Knowing the volume/mass of the solder that left the nozzle per second allowed the solder flow to be calculated. This way the flow of the non-wettable nozzles can be calculated for different wave heights, solder temperatures and alloys. The first experiment compared two non-wettable nozzles with one another. One nozzle has a square opening of 4mm and the other square 6mm. Both nozzles are made of the same stainless-steel material and 3D-printed. The solder alloy is Sn3.0Ag0.5Cu and the solder pot, solder level and temperature are identical. The solder temperature was 300°C. The nitrogen supplied was 17 LPM. The nitrogen was not heated. The frequency n corresponds to a wave height of 3mm. The electro-magnetic pump frequency was varied from n-2 to n+2 [Hz], see Figure 4. For the non-wettable square 4mm nozzle the behavior of the solder for different temperatures was investigated. At higher temperatures solder becomes more fluid but this did not show up in the frequency settings for the pump. A design of experiment with Sn3.0Ag0.5Cu at different frequencies and temperatures showed the impact on wave height. The R-sq = 99% which indicates that the experiment data is consistent. Both solder temperature and electro-magnetic pump frequency have significant impact on the wave height (p-Value < 0.05), see Figure 5. Due to more fluid solder, the average wave height is 0.1mm lower for every 10°C solder temperature increase for a non-wet-table 4mm square nozzle. The laser wave height measurement takes control of solder temperature changes, solder level and coil temperature changes of the electro-magnetic pump to guarantee a consistent wave height in a production process, see Figure 6. For the same nozzle the flow of the Sn3.0Ag0.5Cu solder was
also measured. The solder was collected in a bin. The sample time was limited to a few seconds to minimize the impact of the solder level that decreases when solder is taken out of the pot. The R-sq = 99%. Both factors have a significant impact on solder flow. Due to the higher temperature of the solder the metal becomes more mobile. The efficiency of the electro-mag-netic pump is also influenced by the temperature. As a result, the mass of the solder that flows at 320°C through the nozzle is 30% less than the same pump settings at 260°C. For a wave height of 3.0mm the flow is about 0.15 LPM. The solder pot contains about 8 kg solder ~ 1 liter. So, every 6 minutes all the solder flows through the wave. Solder temperature and pump-frequency have impact on wave height. The other parameter that has impact is the alloy. Different viscosity, fluidity, melting ranges all influence the final shape of the solder wave. Five different lead-free alloys are compared in this experiment. All in the same machine, solder pot, pump, and solder nozzle, see Table
1.For this experiment a non-wettable square 6mm nozzle was mounted in the solder pot. The solder temperature was set at 300°C for all alloys. The next graph shows the impact of the
pump-frequency. All alloys are more sensitive to a frequency e than the Sn3.0Ag0.5Cu. The maximum difference is about 0.5mm for the same pump frequency for this nozzle, see Figure
7.Until here all experiments were done with non-wettable nozzles. The advantage of these nozzles is the solder flow can be collected to measure the flow rate of the solder. In production
processes many lines do apply wettable nozzles. The major reasonis tha t the solder is flowing the same in all directions so the moving of the robot is not significant. For a non-wettable nozzle,the robot needs a rotation, for wettable not. Before investigating the thermal aspects of the different nozzles, the wave height is compared between a non-wettable 6mm square nozzle versus a wettable cylindric 6mm round nozzle, see Figure
8. Wettable nozzles are less sensitive to a change in pump-fre-quency. A higher frequency is needed, however all within the range of the capability of the electro-magnetic pump system and
the requirements for a robust selective soldering process.Wave height and flow of the solder have a big influence on the hole fill. These factors determine the amount of energy that a wave is capable of transferring into the solder joint.This study tries to define the relation between solder nozzle
dimensions, flow rate, and solder temperature on the energy that is conducted to the assembly. To determine the amount of energy a nozzle transfers to a solder joint a Cu test coupon
was defined. The test vehicles used are massive Cu, one with a thickness of 1.6mm and a second of 2.5mm. The specific heat of Cu is 380 J/kg*K. The Cu coupons dimensions are 20x20x1.6mm = 0.64 cm³. The density of Cu is 8.96 g/cm³. The mass of the coupon is 8.96 * 0.64 = 5.7344 g. The energy required to heat up the coupon 1°C is (5.7344/1000) * 380 = 2.18 J/K. In the middle of the coupon, one pin square 0.5mm NiAu plated was placed to have contact with the solder. The protrusion length is 1.40mm, see Figure 9. On the solder destination side of the coupon there are
two thermocouples mounted that record the temperature on the topside. Both are placed 5.0mm out of the corners of the coupon. The average of both thermocouples was used to do the calculations. The solder contact area was calculated with software. The contact area for the non-wetted 6mm nozzle is 67.2mm² and the width is 7.6mm maximum. For the wettable nozzle 6mm the contact area is also 67.2mm² but the width is wider 9.1mm. Temperatures were recorded and converted into energy, see Figure 10. The data shows a better heat transfer for the non-wettable nozzle. The contact areas are equal and the heat transfer for both nozzles are very close. The other interesting observation is that the 4.0mm wave height supplies almost twice the amount of energy to the solder joint. The graph also shows that after about 8 seconds (defined as the maximum dip time by some companies) the energy increases much slower. The wetting of the solder joint should be finished after 8 seconds otherwise it will most likely not come. For the non-wettable nozzle range, the flow, contact area of solder with copper coupon, the calculated solder speed at exit of
the nozzle and transferred energy after 5 seconds contact time are listed in table 2. The solder speed is calculated Q = v solder * AQ = Flow of solder v solder = Speed of solder at exit of the nozzleA = Surface area of the nozzle opening Drag soldering and heat transfer nozzles The point-to-point solder process is mainly a drag soldering process. The nozzles are dragged under the component to be soldered and critical factors are drag speed (defines the contact time), wave height, and solder temperature. The point-to-point solder process is very time consuming, and the drag speed should be high to minimize cycle time. The nozzle with the highest heat transfer will give a sufficient hole fill in the shortest time. To check the performance of the different nozzles thermal couples were placed on the topside pad of a barrel. One barrel with a 4 layer connecter to the barrel and one with only 2 layers. The one with 4 layers had a square topside pad. Board finish was NiAu. Solder temperature was 300°C and the alloy was Sn3.0Ag0.5Cu. The flux was a rosin ROL0 with 10% solids, see Figure 11. The test was repeated for different nozzles.
The objective was to see if there is a correlation between
the calculated energy with the Cu coupon test and the actual soldered joints, see Figure 12. The thermal profiles give maximum temperatures measured on the pad, and the actual contact time with the solder was determined. The component is a single row pin connector with 2.54mm pitch, see Table 3. The drag speed was 3mm/s. The heat transfer is critical for the temperature (and thus hole filling) on the pad with the 4 layers. The next graph shows the correlation between the energy of a
wave and the final temperature on thermocouple 1. To have the complete pad filled with solder the TC1 temperature should at least be higher than 217 °C, see Figure 13. Two of the outliers are due to the wettable 12mm nozzle. This nozzle has a special design to keep the concave shape of the solder wave stable. A cross of metal is inside the nozzle. Thisinf luences the temperature and performance of the nozzle. This nozzle design is very effective for three row pin connectors, see Figure 14.
New generation different type of dip solder nozzles nozzles.
Several trends are driving new types of solder nozzles. The main driver in selective soldering is cycle time reduction. Further, high thermal mass components and fine pitch are trends
that ask for review of available nozzles. The introduction of 3D printing also makes it possible to design shapes that were not feasible with conventional machinery. Four different designs of dip nozzles are compared:
1. Standard static nozzle with side holes
2. Laminar overflowing 3D printed nozzle
3. Turbulent wave nozzle 3D printed
4. Internal overflowing nozzle
For all four nozzles the Cu coupon test was done to define their ability to transfer heat into the solder joint. The same with the point-to-point nozzles the frequency of the pump was
defined for 3.0mm and 4.0mm wave height, see Table 4.
Point-to-point or dip soldering
Drag soldering has cycle time limitations. With multiple solder pots in one machine the point-to-point can be combined with dipping into larger nozzles that are implemented in identical pots. The next experiments were done to see how long dipping is required to get the same amount of energy into a solder joint that is point-to-point soldered, see Figure 15. A static dip nozzle 16x9mm with side flow holes (to maintain a solder flow and provide more heat) was used to solder the same pins of thermocouple 1 and 2 in a dip soldering process with identical solder pot with electro-magnetic pump. The wave height was like the point-to-point nozzles., see Figure 16. Both nozzles require the same amount of free space on the bottom side of the assembly. The width of the solder is 8mm for both. In case of a component of 5 pins with 2.54mm distance the cycle time for drag soldering is 5 seconds which is also the dip time. Due to the dipping all leads will guide solder to the solder destination side and transfer heat. The high mass barrel will therefore be heated up faster as with a drag soldering process as shown in the heating curve. For high thermal mass leads a dip process is preferred. Special dip nozzles are designed to improve the heat transfer.The internal overflow nozzle is a wave that is overflowing over a dam like in wave soldering but all the solder remains within the rim to avoid the risk for washing off small surround-
ing SMD components, see Figure 17. To compare the performance of a dip soldering process, a second experiment was setup. A thermocouple was mounted at the edge of a square pad. A double row pin connector (5×2 pins pitch 2.54mm) is soldered. The next figure shows a schematic of the soldering, see Figure 18. All recorded temperatures for the point-to-point soldering are
listed in the table. The Non-Wettable nozzle square 4mm was not used since it is too small for a double row 2.54mm pitch pin
header. The cycle time for a drag process with a speed of 3mm/s is the same as dipping for 5 seconds, see Table 5. From the Non-Wettable 8x6mm nozzle at a wave height of 3.0mm the picture was made to show the hole filling. Although the temperature of the thermocouple TC3 wasonly 18 0.3°C the picture shows that the solder covers almost 100% of the topside pad, see Figure 19.
The TC3 temperature was not above the liquidus of the Sn3.0Ag0.5Cu alloy for any of the runs. The test was repeated for a dip solder process. For 5 seconds dip the results are listed in
Table 6. The main difference is that the TC3 temperatures for dipping are significant higher. The average temperature of TC3 for dipping is 206°C and for dragging 191°C. The data from the Cu
coupon test the correlation between the heat transfer test and the soldering of TC3 is compared, see Figure 20. According to this calculation a full pad coverage with solder will be achieved for all settings >289 Joule after 5 seconds heating on the Cu coupon. This only counts for the dip process. For 4.0mm wave height the dip time for the internal overflowing nozzle should be 4 seconds or longer to meet this requirement.
Discussion dip soldering
Dip nozzles modified to be used in a solder pot with electro-magnetic pump can transfer enough heat to get proper hole
fill. For connectors with many pins this would reduce the cycle time. The more pins on the connector the more benefit with respect to cycle time, see Figure 21. The dip nozzle with the most heat transfer is the internal overflowing nozzle. This nozzle has a high volume of solder flowing which continuously supplies hot solder by the electro-magnetic pump. The supply of hot solder introduces a lot of heat into the solder joint. The solder stays within the nozzle avoiding risk for damage of surrounding components or other heat sensitive materials, see Figure 22.
This is a method to get enough heat to solder the heavy metal leads of transformers or other devices as earlier mentioned. This nozzle is less capable for soldering fine-pitch components since
it has no de-bridging device like the laminar or turbulent dip nozzles can have.
Design of Experiment hole fill optimization
This experiment should help to optimize hole filling during selective soldering process. The three most significant machine parameters that influence the wetting are included in the experiment:
1. Contact time [s]
2. Solder temperature [°C]
3. Wave height [mm]
The experiment is done for non-wettable and wettable nozzles in point-to-point soldering and for the internaloverflo wing nozzle for dip soldering, see Table 7. The parameters contain the upper and lower settings of a typical selective soldering process. The minimum solder temperature is about 270 °C for Sn3.0Ag0.5Cu and should not be higher than 330 °C. The maximum contact time should not be higher than 8 seconds to avoid material damage. A Cu test coupon 20x20x3.0mm was used to measure the energy for the different conditions. For the internal overflowing nozzle, the energy transferred to the coupon for the different settings is listed in Figure 23. The results are as expected. The contact time has 58% impact, the solder temperature 35% and the wave height only 7%. The nozzle is used to dip the board with thermocouple TC3 for both 300 and 330 °C. The temperature and derivative of the temperature are plotted in Figure 24.
For the drag soldering the Cu test coupon was 20x20x1.6mm. The same test was done to define the energy transfer. The nozzle was placed 2 mm under the coupon and the temperatures were recorded by the two thermocouples 5 mm from the edges. The average temperatures were used to calculate the amount of energy. After this the nozzles also soldered the board with TC3 to enable a comparison with the dip soldering. The drag speed
of this process was 5 mm/s. The distance to bottom side of the board was 2.0 mm, see Figure 25 a/b. In a dip process is the contact time more significant. In apoint-to-point selective oldering process is the wave height more important. The flow of the non-wettable nozzle has a better heat transfer. The average difference is about 30%, see Figure 26 a/b. Compared to dip soldering the point-to-point has a faster wetting. The flow behavior of the solder may influence this. There is only a sufficient wetting at higher temperatures. For the lower solder temperatures, the contact time is not long enough to get a sufficient hole fill, see Table 8. For the non-wettable nozzle wave height 4 mm and drag speed 5 mm/s the solder joint has a perfect topside fillet as
shown in Figure 27.
Discussion
Both processes are capable to get the topside complete covered with solder. The dip soldering is less fast in wetting. Most likely to the static dip process. However, a dip nozzle has more
energy and is therefore able to have an acceptable solder coverage after 5 seconds. The point-to-point solder process is also able to do this. At a drag speed of 5 mm/s similar results as with
dipping for 5 seconds. For this connector the cycle times will be identical. If the components have more leads the dip soldering cycle time will be faster.
Conclusions
Solder machine suppliers respond to the changing demand for manufacturing circuit board assemblies. The trend towards more pin-in-paste reflow leaves a limited number of throughhole components that can’t be reflowed or for which SMD alternatives are not strong enough. Soldering leads of high thermal mass devices needed for power and electric cars require techniques that can apply a lot of heat in the solder joint. Selective soldering machines have multiple solder pots with different nozzles and solder temperatures if needed to solder these devices and minimize the cycle time. Instead of multiple point-to-point solder pots with small nozzles the machines will also have dip applications with wider and dedicated tooling to
meet cycle time requirements and good hole filling. For fine-pitch applications dip soldering nozzles are less efficient because of the bridging risk. For those small devices small nozzles with de-bridging tools are a better solution. For components that are sensitive to bridging, nitrogen gas knives after soldering can de-bridge. For dip nozzles dedicated screens or de-bridging knives solve the problem. This paper showed how the flow of solder of non-wettable nozzles can be measured and introduced a method to define the thermal performance of a solder nozzle process.
