From a young age, I’ve been fascinated by computers. I remember tearing open computers in middle school and marveling at motherboards. Looking at the pin layout on a CPU and comparing it with the socket and connecting all of those pins throughout the motherboard seemed like some sort of engineering black magic.
Have you ever wondered how your PC’s motherboard accurately routes all of the microprocessor pins through the PCB? Enter the ball grid array (BGA). This layout of contact pads makes it possible to route high pin count components to other devices on the PCB. Without BGAs, individual pins would need to be soldered to traces manually. Instead, BGAs make it easy to quickly solder high pin count components to your PCB.
SMD vs. NSMD BGA Pads
There are two types of pads that can be used to design BGAs for your PCB. The solder mask defined (SMD) pad coats the edge of the pad with a small amount of solder mask. This effectively reduces the pad size and ensures that the solder ball sits atop the pad. This thin ring of solder mask elevates tends to elevate the solder ball so that its curvature sits slightly above the solder mask.
There are two other advantages of using SMD pads. First, the opening in the mask creates a channel for each ball on the BGA to align with while the pad during soldering. Second, the overlapping mask helps prevent the pads from lifting off the PCB due to thermal or mechanical stress.
Contrast this with the non-solder mask defined (NSMD) pad. NSMD pads are placed at the same elevation as the surrounding solder mask. The edge of the pad remains exposed, and a small gap is typically left between the edge of the NSMD pad and the solder mask. This gap allows the molten solder ball to collapse onto the pad, which creates a more reliable connection and can help prevent solder from flowing off the edge of the pad.
BGA on yellow PCB with HDI routing
The rule of thumb when placing BGA pads is to set the pad diameter on the PCB to about 80% of the BGA ball size. With SMD pads, you can set the two pads to the same size, and then you can reduce the exposed pad size on the PCB with the appropriate amount of solder mask. Reducing the pad size creates more room between adjacent pads and makes trace routing easier.
Many designers recommend using the NSMD pad style whenever possible due to its improved solderability and pad accuracy. However, different component manufacturers will recommend the best footprint for their products. Always check your component datasheets before placing either type of BGA pad in your PCB layout.
BGA Escape Routing
Escape routing is used to connect the BGA pads to external traces, which in turn connect to other components on the PCB. Routing out of the BGA will typically require multiple layers in order to fit all of the required traces. One PCB layer is sufficient to route a two row thick square at the edge of the BGA. The next two row square deeper into the BGA requires its own signal layer. As you continue to move into the BGA, this pattern repeats, and more signal layers must be added to the PCB.
In BGAa with high pin count and fine pitch, the trace width may need to be adjusted as you route into BGAs. This technique is called “necking,” and changing the width mid-trace can create some signal integrity problems. Changing the trace width mid-trace creates an impedance discontinuity, resulting in signal reflection. In low-speed or low-frequency devices, this effect may be negligible, especially if traces are short enough that they do function as transmission lines.
Surface on a PCB
They typical escape routing method for inner groups of pads is called dog bone fanout. This strategy is suitable when the BGA pitch is larger than 0.5 mm to 0.75 mm. Groups of pads form a square array, and a via is placed at the center of the 4 adjacent pads. This allows a trace from one pad to pass to another signal layer.
The dog bone pattern you create in each quadrant of the BGA should be symmetric under rotation. In other words, if you were to rotate the entire BGA and its dog bone traces by 90 degrees, you would maintain the same layout. This technique is important as it ensures that neighboring traces and vias can be used for differential routing.
When pin density becomes so high that the pad pitch becomes smaller about 0.5 mm, dog bone fanout can no longer be used. Instead, BGA pads can connect to the inner signal layers using via-in-pad technology. If BGA pads are placed directly on vias, the vias should be filled or plated over in order to prevent the solder ball from wicking into the via. Whether you route away from the BGA using dog bone fanout or vias, both methods can be used with SMD or NSMD BGA pads.
The built-in CAD tools and the ActiveRoute tool in Altium Designer 18.1 makes it easy to define your BGA pad size and fanout strategy. You can download a free trial and find out if Altium Designer is right for you. Talk to an Altium expert today to learn more.
Sign up for pre-release pricing for Altium Designer 19 today.
About the AuthorVisit Website More Content by Altium Designer