|
For some years many process engineers and quality managers have been questioning the benefits of solder paste inspection (SPI). Despite the high level of defects associated with this process, the implementation of SPI inspection has not really been adopted in many SMT lines. Many users still question the cost benefit analysis. Other users consider that the need for SPI, and especially 3D SPI, is critical during the new product introduction (NPI) phase or production ramp-ups, but not of benefit during an established process. For them, either the information provided by the SPI does not bring any relevant product quality improvements or they feel illequipped to leverage such improvements. However, with the advanced process now demanding the use of smaller components such as 01005s, μBGAs or PoP (package-on-package) devices, it is clear that the use of in-process inspection has gained importance in the SMT line and should become a standard feature to ensure quality success. Therefore, there is a need to re-explore the requirement for 3D SPI and consider the fact that volume calculation is not enough to qualify the process. Solder print process
The solder paste print process is potentially very unstable with more variation than any other SMT process. According to studies performed by numerous companies and universities this process varies by up to 60 percent. The reason for this variation is the large number of process parameters involved. It is generally accepted that there are about 40 variables to control. Some of these variables include, but are not limited to; paste type and formulation, environmental conditions, stencil type, stencil thickness, aperture aspect and area ratios, printer type, squeegee, print head technology, print speed, etc. Stencil printer performance is typically quantified by measuring the transfer efficiency (TE) percentage and standard deviation of the paste deposition, where 100 percent would mean that the printed paste profile matches the calculated volume of the aperture opening. It is interesting to note that TE varies on a typical SMT board between 20 to130 percent. Transfer efficiency is generally better with rectangular apertures than square or circular apertures; although, volume variations will exist when printing rectangular apertures in either vertical or horizontal orientations. Vertical apertures will print better and with a higher volume. The worst cases for print efficiency are small squares or round apertures less than 12 mil in diameter. Also of note is how standard deviation increases as TE decreases. A decreasing TE indicates decreasing volume repeatability. It is also interesting to note that some studies show that the volume of paste across an array package may not be as important as the consistency of all the pads in an array. Such that, as long as all the pads have the same amount of paste the joints should be acceptable. However, if a few pads are under printed compared to the others in the array, then poor joints may result. 
Technology standpoint
Like AOI (automated optical inspection), the technology in terms of SPI has not evolved too much over the past decade. Up to now, there are two ways to inspect solder paste, one using the methodology of laser triangulation, and the other one based on Moiré technology. Let us compare these two techniques and examine their respective advantages and disadvantages.
• Laser triangulationWhen laser triangulation technology is combined witha 2D image, the height of the inspected object is given bythe deviation of the laser trace (figures 1 and 2).
The drawbacks of this method are mainly linked tothe lack of resolution leading to poor accuracy. Moreover,having only one source (one laser), the exact value of thevolume cannot be calculated. This phenomenon called“shadow effect” is due to the geometrical layout of thesystem. The combination of single angled laser beam andcamera may produce a blind spot on the opposite side ofthe solder deposit to the laser (figure 2). 
In this area, the solder paste deposit is seen as square, as shown on figure 2, and all the quantity of paste contained in this hidden volume (red zone) is not included in the measured volume. The error in the measurement value varies from shape to shape but can be up to 48 percent in the case of round apertures. Using a laser to profile solder paste on a PCB is also subject to PCB color or finish variations.It requires a lot of program maintenance when facing such variations on the production floor. • Moiré methodology Moiré topography is a method of tri-dimensional measurement by phase modulation. In this particular method, the lines projected on the object are modulated interference fringes, and by moving the observation grating, the height and volume of the object can be measured. The drawbacks of the Moiré methodology are the lack of depth of (FOV) field of view (more or less compensated by z movement), the influence of noise and vibration in the grating process, and the cycle time. Moiré methodology is also subject to “shadow effect” if only one source of grating is used. Often, most of the systems using Moiré have two modes of operation, providing speed of operation at reduced accuracy using one source and providing better accuracy at slower speed using two sources. 
With Moiré technology, the inspection is made of a sequence of FOV with individual z-adjustment (the focus is made on average z position in the FOV). From one FOV to the next, covering the same array (BGA type), the references will not be the same and it may result in some differences in term of height or volume measurements. This may produce an unreliable indicator of solder paste quality. Combining the best of both worlds
Recently, a new technology, called Flying Absolute Height Profilometry (FAHP), combining the best features of both methods, has appeared offering an improved solution for 3D SPI in term of accuracy, speed and depth of FOV. FAHP is a scanning solution using dual Moiré methodology. In other words, the board is scanned with a top down camera using two structured light sources (from each side). The resulting image facilitates the calculation of both height and volume of paste eliminating the shadoweffect. This technology, with a 6mm depth of FOV, allows accurate measurement of all types of PCBs, even compensating for high warpage values. During the scanning process, the system is able to measure the PCB profile itself and compensate for the warpage over the whole board (figure 5). 
The slope compensation (or warpage compensation), used at the pad level, provides higher height and volume accuracy (figures 6 and 7). Using “on-the-fly” measurement technology facilitates a continuous measurement across the PCB – the height is referenced around the perimeter of the pad without the need for z-adjustment thereby eliminating potential errors. 
With higher reliability, less dependency on board design or color, 3D SPI, as a measurement system for small devices, will became a useful tool to control the quality of the entire SMT process. In the 3D SPI world, in order to achieve accuracy and repeatability, the measurement system has to eliminate depth of FOV and warpage problems. Getting more from SPI
Today, if most of the engineers or managers are not convinced about the value of 3D SPI, it is surely because the current solutions are not bringing them any valuable quality data. In other words, 3D SPI systems are checking for surface, height or volume and are mainly qualifying the quality of the deposits or the efficiency of the stencil. With wide process tolerances, volume or height measurements do not appear as key drivers in reflow process and joint quality. For example, considering an array component, BGA type, with 105 connections (figure 8), if the process window is set as [40%, 160%] for height and volume measurements, one may end up with nearly 100 percent difference between the maximum and the minimum values in the same array. 
Considering height values of the 105 pads and grouping them on the same graph, the result should be a Gaussian curve centered on the average value and spread within 3 sigma. If the sigma value is low (narrow Gaussian curve), the solder paste distribution will be uniform through the array. Then the “landing area” provided to the component will be seen as flat and will be considered as good for deviceplacement (figure 9). 
If the distribution is really narrow—around an average value of 110 percent with a few pads (2 or 3) with a very low value (50 percent)—bad joints or opens may occur whenthe component is placed and reflowed (figure 10). 
In this case, all the pads are individually good within the process window but the risk of poor quality is real. With such a difference between deposits, the contacts will not bemade evenly on the entire array when the device is placed and pressed on the board (landing area). Thus, there is apossibility that opens may occur in some areas. The samephenomenon can be observed for QFP or SO type deviceswhen the solder paste is not equally printed. In order to beable to compile these results, the 3D SPI system has to bevery accurate and should not depend on the measurementzone (warpage compensation). In the case of very small devices such as 0201s or 01005s, besides the height and the volume, the shape of the deposit has to be considered as an important driver for the final quality of the product. When placing small devices, the contact surface between component and paste has to be maximized by the mechanical pressure of the nozzle. At this stage, the shape of the deposits is a key mechanical factor and will affect the quality of the component placement or solder joint. The shape has to be controlled and flagged as a warning when the component is not placed. By combining solder paste and placement inspection, the system will be able to qualify the quality of the overall process before reflow. The future of 3D SPI in process control
Leveraging software tools such as SPC or closed loop control, the future lies in integrated systems with the ability to manage the quality of the line from a process standpoint. In practice, more and more engineers are using SPC or managing data from inspection tools to control their processes and tighten the process window. The current limitation of this approach is lack of consolidation of overall results gathered from different inspection equipment and tests. Acknowledging that 3D SPI brings much more than height and volume measurements will see such systems become a standard tool in the SMT process. Today, the link between solder paste volume and joint quality on an individual pad has not been proven. However, by considering a clear link between the quality of paste deposit at the component level and the quality of the soldering process, users will realize that the value of 3D SPI systems is worth a lot more than what many have considered them to be. Pad-to-pad volume calculation or shape differentiator, combined with pre-reflow placement inspection, can guarantee the soldering process of the production line without any use of x-ray systems which are costly and time consuming. Driving process control and quality by controlling every step of the process is not impossible but requires the ability to measure the right variables at the right process stage and consolidate them in such a manner as to provide the relevant information. The future lies in the use of the right level of inspection at the right stage and in the combination of the results. | 
