What drives them and how can design teams optimize the process?

In my last article, I discussed the changing role of OEM product development engineering teams (Circuits Assembly, February 2009). OEMs’ downsizing of internal engineering resources, combined with continued requirements for faster design cycles that incorporate more challenging technology, is driving changes in the types of prototyping support required. Here are five trends we see:

Greater need for support of open architecture platforms. Semiconductor manufacturers are trying to address OEM engineering team resource constraints by facilitating the development of single-board solutions that incorporate their processors. One such example is the Beagle Board (beagleboard.org), a low cost, high performance, low power Texas Instruments OMAP3 processor-based platform designed by the BeagleBoard.org community.

The Beagle Board is designed to be either a development system or an out-of-the-box platform for embedded applications such as firewalls, video distribution systems, multimedia picture frames and robotics. A standard platform can cut development time and compensate for the challenges associated with manufacturing the OMAP processor. The processor is a 0.4 mm pitch BGA in a package-on-package (PoP) form factor that accepts a 0.5 mm pitch BGA memory chip on top. The combination of double-pitch BGAs and the microvias found within the board’s layout can make production more complex. Since the manufacturing technology may not be well understood by the OEM engineering team or its prototype build site, having a pre-built platform and open-source design files as options will permit much faster adoption of such advanced components as the PoP OMAP processor. While PoP is well deployed in the higher volume EMS community, it is not necessarily widely used in all standalone prototyping facilities. But we can expect the popularity of these semiconductor manufacturer-driven solutions to continue to drive enhanced support capabilities in the prototyping community.

A wider range of service options to support OEM needs relative to support demands in their end-markets. A survey of our customers found that more than 30% were purchasing quickturn prototyping services to support low-volume/short-run production needs. These are customers that were going through a quickturn prototype process because they found it preferable to trying to outsource a short-run project in a traditional EMS environment. Typically, they were looking for a supplier that would be fast and flexible, and willing to support a project that might have a six-month or less commitment. Project volumes usually didn’t support expensive tooling or non-recurring engineering (NRE) cost. They found traditional EMS new product introduction (NPI) processes too cumbersome for a short-run project, but in many cases, the technology and support requirements exceeded the capabilities of most job shops.  

The market drivers were typically embedded systems products. Robotics and industrial control products in that application segment often have beta test requirements or short-term contracts. We found offering a lower-cost hybrid service with simplified project launch and a longer turn time than the quickturn service satisfied the needs of this market. However, expect the trend of customers specifying exactly what mix of services they want to purchase to continue in the prototyping and short run realm, since they are often pressured by end-customers to do short-term work as well.
Continued streamlining of the prototype process. As mentioned in my previous article, OEM engineering teams find the process of sourcing prototypes to be overly cumbersome. We’ve found value focusing on two parts of that issue. The first area is customer service. Some companies want an easy web-based ordering system, and others want to place an order over the phone.

At the same time, the key to offering customers more choices has also involved understanding key areas of prototyping need segmentation. One obvious division is consigned vs. turnkey prototype orders. However, project complexity also can be a point of market segmentation. Companies and hobbyists with simple needs want simple solutions such as online quote calculators and integrated prototype/PCB fabrication online ordering. In our case, we’ve teamed with PWB fabricator Sunstone Circuits (sunstone.com) to permit customers with simplified requirements to one-stop shop through either company. Expect to see the trend of easier ordering options and supply base teaming to continue, as downsized teams look to take cost and complexity out of their processes.

Continuing requirements for new combinations of engineering disciplines. While engineers have trained in specific disciplines, many products have experienced significant technology convergence. For example, wireless is analog. Digital engineers may have been told in school that analog was a dying technology, but in mixed signal applications, there is a requirement for skills in both areas. Point of load power and reprogrammable memory use also drive the need for expanded engineering skills. While tapping the skills of your prototype partner is one solution, design engineers should also recognize that engineering degrees are the first step in the journey. Today’s technology requires continued skills enhancement.

Inefficiencies in prototype design driven by OEM resource constraints. Downsizing is here to stay. Today’s engineering managers have two choices:
They can continue to juggle shrinking resources and hope for the best, or they can strategically map processes for supporting company product development needs with reduced resources. Tactically, that means evaluating suppliers not only for their ability to build prototypes, but also for their ability to team in supporting the product development process. Flexibility and scalable resources are two key areas to consider in this evaluation process. Additionally, it is important to determine what “free” resources your supply base can offer. Are there design guidelines that can be shared and built into internal layout processes? Are there educational resources that discuss common layout mistakes in newer technologies that can help your team as projects increase in technological complexity? Find ways to integrate your suppliers’ resources into the gaps in your team. Companies that address trends will lead the pack in competitiveness.  

Duane Benson is marketing manager at Screaming Circuits (screamingcircuits.com); dbenson@screamingcircuits.com. For additional tips, visit his blog at http://blog.screamingcircuits.com.

Complex, BGA-laden PCBs demand one-of-a-kind profiles.

PCBs laden with BGAs, some on both the component and solder sides of the board, are a challenge to thermally profile. And not only are PCBs increasingly populated with challenging BGA packages, but they are becoming more dense and smaller to comply with portable and wireless applications. Add to that the vast number of board sizes.

All of this demands special attention to thermal profiling, regardless of the end-application. In fact, each board requires a unique thermal profile. This also means different profiles for eutectic SnPb and Pb-free solder pastes. Special care must be taken in creating these profiles, as Pb-free solders require a higher peak temperature compared to eutectic solder.

The first step in creating a thermal profile for a specific PCB is to obtain a solder sample board. This board is used as the basis for creating a unique profile. Next, obtain the type of solder paste to be used. Special care should be applied here to differentiate between leaded/eutectic and Pb-free solder paste.

The solder paste manufacturer plays a vital role in the overall scheme for creating the correct profile. Here, it is imperative the manufacturer’s solder paste specification be used to set up a correct thermal profile. The manufacturer provides general guidelines and suggestions for optimum temperature settings relating to its particular solder paste.

The Setup

For a double-sided, BGA-populated board, first create a profile for the bottom or solder side of the board. Thermocouples are connected on specific BGAs. If BGAs are located at multiple locations, thermocouples are placed at each location.

After proper locations are determined, thermocouples are placed on those pad areas where BGA components go. Thermocouples are placed on the edge and in the middle of the BGA to achieve close or exact thermal readings from both areas. Thermocouple placement also applies to smaller BGA packages as well. However, while two thermocouples are used for larger BGA packages, a single thermocouple can be used for smaller BGAs. It is ideal to place a thermocouple at the edge of the board to verify that this area of the PCB is not overheated.

Aluminum tape is placed on each thermocouple (Figure 1). Dummy BGAs, similar to actual components, are then placed on top of the aluminum tape, which measures the temperature of that particular area as the board goes through the different reflow oven zones.

Temperature settings assigned to the oven are largely based on experience. Here, similar temperature settings from previous builds are used. The conveyor speed and temperature settings are entered into the oven and thermal profiler software.

Next, the board is ready to go through the oven, meaning the profiler is turned on, and the PCB and profiler are run into the oven. It usually takes five to six minutes to pass through an eight-zone oven. When the board exits, it is plugged into the profiling station (Figure 2). This station gathers the board’s temperature settings and provides actual readings of each location where a thermocouple is placed. Those readings provide peak temperature, soak time and reflow time for each component.

Actual temperatures that each component has undergone are shown on a monitor. That visual also displays the readout from that particular profile. A green reading shows the profile is within the solder paste manufacturer’s specifications. A red reading shows the profile is outside those specifications. A projected or suggested good setting for a particular profile is given underneath the actual readings (Figure 3). The green reading shows that the profile is within manufacturer’s specifications. The red reading, in this case, the soak time, indicates the profile is outside those specifications.

In this example, an eight-zone convection oven is used. The thermal profiler collects readings as it goes from the first zone on to the last cooling zone of the oven. Those readings are then compared to the solder paste manufacturer’s recommended application settings. As the board exits the oven, the thermal profiler is attached to the profiling workstation and thermal readings are checked. Those readings are then compared to the solder paste manufacturer’s suggested reflow profile guidelines.

Some adjustments may be necessary if the readings don’t closely match. In some cases, a number of adjustments are required. The profile is then run again for that particular board using the profiler’s predicted temperature settings. Afterward, in most instances, the profile readings are close enough to label it a correct profile.

After the bottom side is processed, the topside of the same board is populated using dummy BGAs and other components using the same procedures as those for the bottom side, including placing thermal couples at specific BGA locations. Then, that particular profile is run and temperature readings are taken across the board, from the edges, and from bottom-side BGA components. Again, the objective is to get as close to or exactly the same temperature readings as though the actual assembly is being processed.

The goal here is to have an appropriate temperature profile for the bottom (solder) side, as well as the top component side. This profile depends on the component density on each side of the board, number of board layers, and amount of copper used in the layer stackup.

After the board exits the reflow oven, data are collected from the thermal profiler. The objective is to get the temperature readings as close to the paste manufacturer’s specifications as possible. If the readings aren’t correct after the first run, a second profile run is performed to get exact readings or as close as possible to the paste manufacturer’s suggested settings.

It’s crucial to concentrate on BGA component temperature readings for a large, complex, and heavily populated board. However, it is equally as important to monitor temperature readings from other components, as well as from the board. This is especially true for such temperature-sensitive devices as QFNs and CSPs.

Temperature readings are taken on these particular devices when a profile is being created. The reason is when higher temperature settings are applied on larger components, it’s important to ensure smaller components don’t overheat. Applying too much heat causes component failure or causes the solder paste to exceed manufacturer’s required temperature readings. That’s why it is important to efficiently and strategically spread thermocouples across the board to achieve maximum reading accuracy. This includes monitoring the board middle and corners.

It’s challenging to obtain exact readings across the board. The best approach when performing profiling is to get the smallest deviation relating to the temperature readings. A temperature deviation of 10ºC or less from the different areas on the board is acceptable. Even if 10ºC is exceeded, the deviation is okay, provided it is within the paste manufacturer’s specifications.

The rule-of-thumb is to run the first good board through the oven and conduct an inspection prior to running the remaining assemblies. The first article usually makes clear whether all components on the board are properly reflowed. Quality control will not only inspect each individual component to ensure a good solder reflow, but will also check for other process defects such as tombstoning and misaligned components. Sometimes, the first board provides indications that certain profile adjustments are necessary and should be made immediately. The second board going through the oven also gets close inspection to ensure standard solderability requirements are met.

Side-Stepping Pitfalls

Developing a correct thermal profile calls for close cooperation among all parties. Miscommunication can lead to costly mistakes. For example, there are cases in which a project specification calls for SnPb or eutectic solder paste, but the assembler receives BGA components with Pb-free solder balls. This mismatch can pose a major problem.

Pb-free solder balls require higher temperatures for reflow. Using a SnPb profile in this instance would create a cold solder condition because the manufacturer’s recommended temperature setting is not applied. Moreover, reliability would be compromised. At times, BGA microcracks are not visible to QC or even to x-ray, meaning they are latent problems. Yet, over time those defects could emerge as field failures. To avoid these and related issues, take care to ensure the correct solder paste and processes are used for each PCB.

Other issues that can come up are solder balls and PWB delamination. Solder balls typically occur when the board undergoes an excessive heating rate as it passes through the oven. An excessively long reflow cycle can cause solder balls to form throughout the board. Also, excessive heat can cause delamination issues on the PCBs and compromise board reliability (Figure 4). Tombstoning, a defect in which one side of the component is pulled and causes the other end to stand up, is another issue caused by excessive heat (Figure 5).

The Role of Thermocouples

A thermal profiler can accommodate nine or more thermocouples. Usually, about five or six are used on one side for a complex, BGA-laden PCB. But a typical thermal profiler can accommodate the maximum capacity for each profiler, permitting temperature readings from all thermocouples.

The most important consideration is determining where a thermocouple should be used. Definitely, it should be placed on a BGA component. Also, temperature readings on the PCB’s edges must be obtained, and the process engineer needs to ensure no board area is overheated.

After placing thermocouples on BGA components, other thermocouples need to be placed on various ICs, resistors and capacitors. The idea is to spread them throughout the board to collect a sufficient number of readings to ensure the appropriate amount of heat is applied across the board.

What happens if thermocouples are improperly located? Generally, they are to be placed at strategic areas of a board. However, for example, if they are erroneously placed in a board area with few components, obviously the benefits of efficient temperature readings are lost. The only benefit is ensuring that particular area is not overheated.  

Simon Ilustre is a process engineer at NexLogic Technologies Inc. (nexlogic.com); info@nexlogic.com.