Press-fit connectors are widely used in system interconnections. In multi-board systems, most functional subassemblies are connected through press-fit connectors. Recent concerns with copper dissolution associated with rework of solder tail connectors may also encourage more use of press-fit connectors especially for complex products which are assembled with lead-free solder materials. With the increased density of the printed circuit board assemblies (PCBA), the lead pitch has decreased for press-fit connectors and at the same time, the PCB layer count has increased. Technical challenges have emerged due to these factors resulting in defects in the PCBAs. Some defects which encountered were padlifting, crazing and laminate damage. Connection reliability
A press-fit connection is made between a metal lead of the connector which has a compliant section and a copper plated through-hole (PTH) on the PCB. An assembled connector is shown in figure 1. Figure 2 shows the hole array where a press-fit connector will be inserted. Figure 3 shows a cross-section of a wellinserted connector leads in PTH’s. There are three main elements which guarantee a functional and reliable press-fit connection: 1. Proper press-fit lead normal force against the sidewall of the PTH. The press-fit pin connection is reliable due to the normal force exerted on the side walls of the PTH creating a gas-tight seal and electrical connection. 2. The integrity of copper plated through holes in terms of copper thickness, ductility and hole roughness plays an important role in the reliability of the connection between the PCB and the press-fit connector pin. 3. No excess mechanical damage to either the press-fit pin or the PCB PTH wall structure and surrounding laminate material. The insertion process should be controlled to press smoothly, seat the connector and monitor the insertion force during the different phases of connector insertion: Initial force ramp, peak entry force, plateau force and the seatingforce. Monitoring the insertion load curve is an excellent process control method to assure that the connector,PCB and process are all running within specification.Several out-of-spec conditions and defects can bedetected through this method. Typical press-fit defects include bent pins, pad lifting, crazing, delamination, PTH breakout by the press-fit pin, inner plane pull-down, and low retention force. Low normal force results in poor electrical connection and corrosion of the contact surfaces. High normal force results in PTH damage and laminate damage, plastic deformation/breakage of the press-fit lead, etc. Problem approach
Focusing mainly on the crazing as the primary defect and pad lifting as the secondary defect, the cause of crazing was investigated (see figure 7). It has been shown that there is a critical stress value, Kc, below which craze formation will not occur. Craze propagation will continue under continued application of stress to the polymeric material. In approaching the problem there are two assumptions that can be made. One, either the critical stress value was exceeded by a change in connector, process or reduction in PTH size leading to an increase in the level of stress between the connector and the PCB. Or two, the critical stress value for the epoxy in the PCB has been reduced by a change in the PCB. The initial assessment included all three possible sources; the connector, the PCB or the insertion process could be at fault for causing the crazing and pad lifting observed. Incoming connectors were inspected for any obvious abnormalities which may exist. PCBs were inspected for board thickness, finished plated through hole size and the general condition of the PCBs. Insertion and extraction tooling dimensions and function were all verified. Mismatch of connector pin tolerance and PCB hole tolerances can be a major concern. Excessive stress is generated due to this kind of positioning offset. The connector pin positioning tolerance is controlled by the plastic housing. Ultimately, it is directly attributed to plastic molding die set. A loss tolerance can be the result of a worn die. From the PCB side, plated through hole positioning tolerance is controlled by the precision of PCB drilling machine. The PCB dimensions and the insertion and extraction tooling were all checked out with no obvious issue observed. A difference however was noted between the same connector made by two different suppliers (seefigure 8). The edge coining to round the edges of the eye-of-the-needle press-fit pin was much more pronounced on one connector versus the other. The coining ispreformed to better match the radius of the press-fit pinto the radius of the hole to minimize damage that thepin will produce in the hole during either insertion orextraction. Insertion process controls
To attempt to differentiate between these main variables two PCB suppliers were selected, two connectorsuppliers were selected and two insertion press speeds were also used. An MEP 12T electric servo press (seefigure 9) was designed for the press-fit connectorinsertion process. Key variables which are controlledby the press are the minimum and maximum allowedinsertion force and the insertion speed which are bothcontrolled during each step of the operation. Connector insertion termination can be determined by a number of methods including z-stop, increase in slope due to connector housing seating, maximum force reached or a percentage increase in force over the plateau insertion force. Figure 10 shows the initial increase in force as the connector is pressed on top of the PCBA. It also shows the peak insertion force as the lead first enters the PTH, and the plateau force as the lead is sliding further into the hole, as well as the termination of pressing as the slope significantly increases as the housing is seated. Among variables of the connector, PCB and insertion process, there are different possibilities of what could cause the increase in crazing. According to M Kitagawa, crazing cannot occur without dilatational stress. By the very nature of press-fit connectors, dilatational stress or stretching stress is always present. However, according to CH Park, threshold activation energy is also required in order for crazing to occur. l = l0exp[-(Q-σv)/kT] Where l is the craze velocity, v is the activation volume, Q is the activation energy, σ is the stress on the polymer, k is Boltzmann’s Constant, T is temperature in degrees Kelvin and l0 is a constant. What caused the change to increase the severity and frequency of crazing? The most obvious answer is within the connector if the insertion force is increased appreciably, this increase can relate to an increase in energy transferred to the PCB and therefore an increase in crazing. However, since the press will not allow a press cycle to proceed if the maximum stress goes beyond acceptable limits and the force curves have not changed, this is not the case. The difference in degree of coining (rounding) edges of the press-fit leads can lead to an increase in localized stress, σ, leading to more craze initiation. With respect to the process, the only variable which is controlled by the process itself is insertion speed. If the insertion speed is decreased, perhaps the stress within the PCB can be absorbed through creep stress relaxation of the polymer rather than craze formation. PCB glass-cloth/epoxy laminate is a complex structure with many factors that can lead to craze formation if they are changed. If the copper PTH thickness is reduced, more energy would be transferred directly into the laminate and could easily exceed the threshold for craze formation. A change in the molecular structure of the epoxy polymer, molecular weight, molecule orientation, degree of polymerization, can lead to a change in the activation energy necessary for the formation of crazes. Micro-voids, poor impregnation, poor adhesion of the epoxy to the fibers, local resin rich areas could lead to a decrease in mechanical strength of the sample and again a decrease in the activation energy necessary for craze formation. The problem at hand is to separate the potential causes and determine the factors which lead to the observed crazes. Conclusions
From experiment results, all three factors - connectors, PCBs and process-played a significant role in the formation of crazes. The process was adjusted to the optimum connector insertion speed. While the connector coining was not the most significant factor, it was a factor and this was communicated with connector supplier so that they could improve the edge coining of the press-fit leads. After the process was optimized the most significant factor was in fact within the PCB itself. The PCB supplier adjusted their process such that the defect rate and severity were reduced to where they are in compliance with the specifications. ------------------------------------------------------------------------------- What is crazing? Often running side by side with pad-lifting, crazing in amorphous glassy polymers such as the resin found in printed circuit boards is defined as where the random coils of the polymers are mechanically strained in one direction. From this strain the random coils start to take on an orientation resulting in a crystalline alignment of the polymer at the point of strain. Crazes are not true fractures but made of fibrils and voids. The alignment of the polymer results in fibrils which take on a cloudy appearance. There is usually a marked boundary between the crazed area and the area which is not damaged. As the strain is increased the fibrils grow longer and eventually break leading to crack formation and delamination of the PCB. Crazing is found at the leading edge of the strain and is a potential precursor to delamination within the printed circuit board. IPC assumes worst case and identifies the typical place crazing occurs. It defines board crazing as an internal condition occurring in the laminated base material in which the glass fibers are separated from the resin. This condition manifests itself in the form of connected white spots or crosses below the surface of the base material and is usually related to mechanically induced stress. The local damage of the PCB structure could cause loss of mechanical seal allowing chemical ingress and degrade the integrity of the PCB. The resulting defects include barrel to internal circuitry separation and growth of metal dendrites. Figure 4 shows a typical example of crazing in a top view looking down at the PCB after a press-fit connector has been removed. As is often done for qualification or for connector rework, the connector needs to be removed for this view of the board to be seen. Figures 5 and 6 show a cross-section of crazingcaused by the insertion of a press-fit connector. ------------------------------------------------------------------------------- Acknowledgement
The authors would like to thank the efforts of Hyman Liang and Suranat Pinvises during this investigation. They would also like to thank the Celestica Song Shan Lake materials lab for their efforts to measure, cross-section and analyze the defects. Originally published in the 2010 SMTA International Conference Proceedings. About the authors
Phil Isaacs and Sven Peng are based in IBM Rochester, MN, USA and Shenzhen, China respectively. Phil can be reached at pisaacs@us.ibm. com. HM Chan, Wai Mun Lee, and Alex Chen are based in Celestica Song Shan Lake, China and Toronto, Canada respectively. Alex can bereached at achen@celestica.com. | 
