X-ray's 2-D vs. 3-D debate: To slice or not to slice?

Escalating pin counts, the use of flip chips and ball-grid arrays (BGAs), and crowded, small PCBs have increased the clamor for test methods that do not require circuit access through a bed of nails. Manual and automated optical inspection (AOI) and metrology techniques such as laser mapping offer some help, but these techniques require that components and PCB features be visible. Today's PCB designs can't make that guarantee. To cope with the need to inspect hidden components and connections, more and more companies are turning to some form of x-ray inspection.

Unfortunately, x-ray inspection is not a single technique but a family of techniques, each of which will perform more or less well in particular manufacturing situations. Achieving success among the x-ray alternatives depends on a PCB's topology, technology, application, and a host of other issues.

X-ray imaging falls into two broad categories: 2-D and 3-D. Two-dimensional (2-D), or transmission, x-ray inspection involves a source shining down perpendicular to the board's surface, thereby "seeing" both sides at the same time. Three-dimensional (3-D) techniques such as laminography and tomosynthesis examine horizontal slices of the board at desired z-axis heights, and can therefore analyze several horizontal layers of a PCB in a single inspection pass. Examining the relative merits of 2-D and 3-D techniques will help you understand where each fits into the arsenal of test, inspection, and quality tools. (Discussions of manual vs. automated systems or laminography vs. tomosynthesis are beyond the scope of this article.)

Because a 2-D x-ray system shines a beam through a PCB, any object in the path of the beam that is relatively opaque to x-rays will show up as a dark shape in the image. Advocates of 2-D contend that because the constructed image includes the entire z-axis depth (or density) for components and similar objects, edge contrast and resolution can be higher than those for images obtained with 3-D techniques. Supporters of 2-D also decry the difficulty of programming 3-D systems and the fact that 3-D systems cost at least half again as much as their 2-D cousins.

Proponents of 3-D argue that today's densely populated two-sided boards preclude the 2-D approach. When objects on the top and bottom sides of a PCB overlap, distinguishing individual features becomes difficult or impossible. The 3-D camp also emphasizes the better fault-coverage statistics and diagnostic capabilities of 3-D x-ray techniques.

There are several points on which 2-D and 3-D supporters agree. 2-D equipment is less expensive, and because there is no z-axis height to consider during inspection, programming can take considerably less time. Yet these factors alone would be insufficient to encourage people to adopt 2-D techniques if the method did not also find adequate numbers of faults.

2-D x-ray offers shorter inspection times than 3-D because 3-D needs to construct multiple images—one per "slice" through a PCB. Proponents therefore recommend it for real-time analysis in high-throughput situations. The most common 2-D application is in the automobile industry, where the hostile electronics environment demands stringent quality measures, and the use of BGAs and flip-chips encourage x-ray methods.

2-D systems also work well on PC motherboards and boards destined for other consumer products. Although these types of boards often contain components on both sides, the types of components and their distribution on the two sides is generally very different. For example, digital packages may populate the "top" board side, while the "bottom" side includes sparsely distributed resistor packs, discretes, and other components whose contacts will not generally line up with those from the top side, except by coincidence. Therefore, looking directly through the board causes little ambiguity or confusion. A 2-D x-ray system can easily distinguish individual features to determine fault conditions. Even when devices on top and bottom occlude each other, the simple design-for-inspection step of staggering the nodes between the two sides will reduce the resulting loss of fault coverage.

Figure 1 shows a typical 2-D x-ray image of a gull-wing component on an automotive PCB. Note that the three end nodes at the top left of the device differ from the nodes around them. The solder did not flow properly on those nodes while the board was in the reflow oven, so the leads will likely not make a clean connection with the pads on the board surface.

Figure 1. A 2-D x-ray image of a gull-wing IC. The three joints at the upper left corner of the IC look different from the others because they did not reflow properly in the oven. (The large dark area represents a component on the other side of the PCB). Courtesy of GenRad.

Even 3-D defects such as voids in solder balls can be detected with transmission equipment. The void appears as a lighter gray area on an otherwise much darker gray image. In fact, 2-D advocates contend that finding solder voids is easier with a 2-D system. Because a 3-D system examines z-axis slices, it must select a profile height that cuts through the void, which requires knowing the location of the board surface and any unevenness in that surface. The 2-D advocates also contend that because of the need to slice the board at a particular z-axis height, a 3-D system cannot find a fault that the programmer doesn't know enough to look for.

For example, on many quad-flat-packs (QFPs), the solder wicks its way up the lead, possibly creating a short closer to the top of the lead than to the board surface. This so-called "high bridge" is easier to detect using 2-D equipment because 3-D systems generally don't examine profiles at that height. Also, RF shields and heat sinks—which may make determining the board surface more difficult—present no impediment to a 2-D x-ray technique that looks right through them.

Figure 2. When a board contains similar components on top and bottom sides (a), the 2-D image (b) has difficulty distinguishing them. (c) 3-D techniques look at only one side at a time. Courtesy of Agilent Technologies.

Also, 3-D x-ray systems can quantitatively measure the volume of the solder joint, determining heel length, fillet height, and other characteristics. These measurements can help ensure that the reflow part of the process is working properly. If these values suggest to you that a process is out of control, you can take steps to correct it. This type of process feedback can maximize yields from production and thereby raise overall yields and overall product quality.

Figure 3. These photographs contrast 2-D and 3-D images of the same ball-grid array. (a) 2-D techniques show the whole ball in each case. (b) The laminographic slice at the board level shows that one of the balls makes insufficient contact. Courtesy of Agilent Technologies.

Manufacturers may choose a 3-D system simply because of its versatility, even though 2-D systems function very well in certain applications. For example, a test operation that works with a variety of PCB designs—such as the test operation at a contract manufacturer—cannot necessarily predict the suitability of 2-D inspections, so it opts for the 3-D alternative.

One factor changing the dynamics of 2-D vs. 3-D x-ray inspection is the growing movement to minimize testing by discouraging the age-old practice of looking for every conceivable fault at every test step. Instead, distributed testing looks for as many faults as possible at each step and identifies them for repair. Then, "downstream" testers don't have to look for those same problems. Adopting this approach frees the test operation to select test steps that may not find every fault, as long as another step in the process will. Because most manufacturers place inspection prior to test, common practice looks for as many faults as possible at inspection, which can permit a manufacturer to reduce its in-circuit test requirements. The manufacturer needs fewer and less-elaborate machines and less-complex test-generation processes, thereby lowering its costs. Proponents of 2-D x-ray techniques contend that, in this scheme, complementing x-ray with in-circuit test considerably reduces the fault-coverage advantage of 3-D x-ray systems.

Some x-ray system manufacturers combine 2-D and 3-D capability on the same machine. This form of distributed test finds as many faults as possible using 2-D inspection, then applies 3-D analysis to those board areas that require it, thereby minimizing the impact of 3-D inspection on test time.

As with all test and inspection issues, no panaceas exist. Neither 2-D nor 3-D x-ray can address all quality problems. To decide between them, you must examine your overall quality goals, the nature of the product, and your manufacturing processes. You should also consider whether your next product will vary substantially from ones you produce now. Large differences in product types may demand different analysis of test and inspection options. Only after you perform a comprehensive analysis of the costs and benefits of each alternative can you select the option that will work best for you.