In recent months, I have encountered a number of high-layer-count PCB designs with high-pin-count BGAs, such as FPGAs with 1 mm pitch contacts, designed with blind and buried via technology. Using this technology, these PCBs have been difficult to design, difficult to manufacture, more expensive than their through-hole counterparts, and difficult to test and troubleshoot. At the same time, PCBs of similar complexity and density were being done in the same number of layers using standard through-hole technology with none of these drawbacks, and at lower overall cost.

Two specific cases were eighteen-layer PCBs used in networking products. The same basic circuit was being manufactured using standard through-hole technology with all of the advantages of low cost, ease of layout, and ease of test. The fabrication drawing consisted of only one page and had a single drill file. At the same time, a second PCB using the blind and buried via approach cost 20 percent more to build, took twice as long to layout, was very difficult to test, and had an eight-page fabrication drawing, four drill files, and still required eighteen layers.

I was curious as to why these designs were being done this way when it was not an improvement over the standard through-hole method normally used for such designs. Upon investigation, I discovered that there are presentations being made at conferences and some classes being taught that claim that combining blind and buried vias on this class of design saves layers by allowing many of the signals to be routed on layer two of the PCB. From practical experience, this has turned out not to be true.

There are places where combining blind and buried via technology serves a very important role. Among these are cell phone PCBs and high-pin-count BGA packages. In these cases, the motivation isn’t to save layers. Rather, it is to make possible designs that would otherwise have no solution. Cell phones, due to their compact size, have fine pitch components mounted on both sides of a single PCB. If standard through-hole technology were employed, component holes from components mounted on one side of the PCB would penetrate mounting pads of components mounted on the other side. In order to avoid this problem, blind vias are used to reach into the second or third layer of the PCB where connections are made to other circuit pins or power rails. This is done on both sides of the PCB connecting to a common core in the center that contains the power distribution networks. The result is a core with four or more layers built with conventional through-hole technology that has built-up layers on both sides, which are connected to the core using blind vias. This is often referred to as build-up technology. It is easy to see that this process will cost more than a like number of layers using standard through-hole technology. This is the price paid for miniaturization.

Dense, high-pin-count BGA packages use blind and buried vias in a build-up process much like that used for cell phone PCBs, but for another reason. The balls or bumps on the BGA die are usually placed on an 8 mil (.203 mm) pitch. This pitch is far too fine to allow through-hole vias. As a result, tiny, blind vias are used to penetrate to layer two and three of the package where very fine traces fan out to the 1mm or 50 mil pitch balls on the bottom of the package that finally interfaces with the main PCB.

In neither of these cases was the motivation to save layers, save design time, or improve testability. In fact, the opposite is usually true. The use of blind and buried vias was to make the product possible at all! Testability is compromised in the case of the cell phone PCB due to the fact that few, if any, of the component pins are accessible for conventional in-circuit testing. As a result, the components used in such a design must contain special test circuits such as boundary scan or JTAG that allow testing from a few test pins that access each IC from a special test connector. When a design is done with through-hole technology, all device pins are accessible from the back side of the PCB making it easy to do in-circuit test, as well as allowing easy attachment of oscilloscope probes during troubleshooting.

After much review, I have determined that those giving the advice to use blind and buried via technology on PCBs where it is not the best solution are well intentioned. However, those who follow this misplaced advice are paying a very high price. When a technology that works in one area is transferred without adequate qualification, the results can be mixed at the very least and disappointing at the worst. Those giving the advice owe it to their audience to make sure that, in their enthusiasm to promote a concept, it fits where it is being offered. Not doing so can result in some very bad end results that are not cost competitive or timely.