Infrared inspection finds unexpected hot spots

The principles behind infrared inspection of printed-circuit boards (PCBs) are relatively straightforward. Simply stated, a faulty circuit produces more or less heat than a good one does. Assuming you can precisely measure the resulting temperature differences, you should be able to compare the temperature map from a board undergoing infrared inspection with one from a known-good board to accurately pinpoint circuit problems.

So why is the technique rarely implemented, especially in mainline production? Chris Bainter, senior science segment engineer at FLIR Systems, said that the biggest reasons are unfamiliarity with the advantages of infrared inspection and a perception (incorrect) that it would prove more expensive than the available alternatives.

“The biggest obstacle that we encounter,” said Bainter, “is that the industry isn't sufficiently aware that infrared inspection can reduce test times and improve product quality, offering a quick return on investment. As a result, you generally find infrared inspection today only on lab benches for design verification and in repair depots.”

Bainter does not suggest that the technique is inexpensive. “There is no denying that the initial investment may seem daunting or even prohibitive. The least expensive infrared cameras cost about $10,000, and more elaborate solutions can approach $70,000 apiece. Still, if you factor in reduced test times and improved inspection accuracy, the return on that investment can be enormous.

“General computer makers already use the approach quite frequently. When a customer sends a product back to the manufacturer for repair, infrared inspection can examine an entire board or system in a 30-second snapshot without bed-of-nails fixtures, simulations, or other expenditures that are unique to a particular assembly. The only 'fixturing' required simply keeps the board from moving during image capture.”

For verifying designs on the R&D lab bench, IR inspection serves as a holistic alternative to mounting a collection of thermocouples on the board at suspected “hot spots.” Thermocouples measure temperature only at individual points. If a board designer fails to anticipate (and therefore neglects to measure) one or more such hot spots, the board may fail at unacceptably high rates either during production or in the field.

Fig. 1  This low-resolution IR image showed a pattern of heat dissipation on a PCB that differed markedly from the predictions of a design simulation. Courtesy of FLIR Systems.

“When we visited their operation, our first image—a fast but low-resolution 320x240 [Figure 1]—made their collective jaws drop. It revealed locations on the board that dissipated much more heat than the points that the simulations had predicted. They told us that in five minutes they had learned more about the board design's heat profile than they had in months or years before that.”

But the best was yet to come. Bainter's crew followed up with a higher-resolution detector that took a 640x512 image, and the customer received another surprise (Figure 2). “The second image had uncovered a previously unknown hot spot in a remote area of the board that no one had even considered before.”

Fig. 2  This higher-resolution IR image of the same PCB as in Figure 1 revealed a “hot spot” in the upper-right corner of the board, where no one had expected any thermal problems. Courtesy of FLIR Systems.

Higher-resolution analysis requires more elaborate equipment. An indium antimonide (InSb—pronounced “insby”) detector can achieve a resolution of 1024x1024 with a smaller pixel size. At the high end of the cost range, these detectors can measure temperature differences below 0.018°C. But to operate, an InSb detector requires cryogenic cooling to a temperature of approximately 77 K, either using a liquid nitrogen pour-fill Dewar flask or a closed-cycle sterling or rotary cryogenic cooler.

One aspect of infrared inspection that differs from techniques like automated optical inspection (AOI) and x-ray inspection is that infrared systems look at a board while it is running. “You can power the board up,” commented Bainter, “and wait until it reaches steady state before performing the inspection, or you can start cold and measure it as the temperature rises. The latter process can be extremely valuable because some components may heat up excessively before settling back into a normal range.

“A cell-phone battery, for example, might run hotter when a call comes in than when it is idle. With good fixturing to keep the board from moving, you can inspect it both cold and under power, then by image subtraction, look for subtle differences between the two.”

Manufacturers generally conduct both an initial inspection and a higher-resolution examination of suspect areas with the same camera, adding magnifiers and other optical accessories to home in on areas of interest. With microbolometer cameras, that second step generally means changing lenses. InSb cameras permit extender rings and spacers that allow the camera to “zoom” in and magnify the area of interest with a single lens. Because lenses often cost $5000 to $10,000, that added cost with the “inexpensive” microbolometer solution sometimes drives manufacturers to migrate toward the higher-end single-lens alternative.

Bainter also explained that infrared inspection offers advantages over more conventional test and inspection steps. “In-circuit test requires mounting the board on a bed-of-nails fixture that by itself can cost more than an entire infrared setup. And that fixture is assembly-specific. IR inspection is often faster, which increases throughput and thereby lowers costs as well. The reduced test time may even permit inspecting every board instead of inspecting only parts of boards or board samples. And infrared systems are generally easier to use and to maintain than more complex test systems.

“Also, infrared cameras don't require visible light to generate a good image. In optical inspection, a passing shadow or other momentary light variation can produce false failures or escapes. Infrared inspection can prove much more consistent because the object being inspected is generating heat—and therefore infrared light—which is not susceptible to visible-light inconsistencies.”

Some manufacturers also offer “smart” IR cameras that can inspect entire boards and automatically determine which areas exceed a user-defined temperature threshold. Upon finding such an area, the camera creates a real-time alarm. It can then output a digital or analog signal, giving feedback directly to a PC, programmable logic controller (PLC), or other in-line controller, thereby flagging potentially bad boards for further testing off-line. This level of automation reduces the board-to-board variability of the inspection results and further improves product quality and reliability.

As the capabilities of infrared equipment continue to grow, the costs continue to decline. Bainter commented, “Depending on your specific inspection requirements, some cameras cost less than the $10,000 threshold and can fit in the palm of your hand. Prices are falling to the point where, within five years, we will even be able to address the low-cost, low-margin consumer market.”