Mislabeled passive components (reels of resistors and capacitors) are surprisingly common in the supply chain. Testing SMT passives first can be the best option.
Once placed on boards, the wrong part is very hard to detect and expensive to correct. We buy reels of SMT passives by the thousands for our contract manufacturing business, usually through major distributors. Several times a year we get a mislabeled reel, so that the true value of the component is different from what the label says. If we don't discover this error before placing the parts, at 5,000 parts per reel, that's a lot of bad boards.
Worse, normal production tests are unlikely to catch this type of error once it occurs. Most surface-mount technology (SMT) passives we use today are 0603 (0.06″ x 0.03″) and smaller, and not individually marked due to their small size. One hundred percent automated optical inspection could confirm that a passive with the right package size is placed in the right location, but its value cannot be determined: a 10-ohm resistor looks exactly like a 10K-ohm resistor to a camera.
In-circuit testing could catch this type of error in perhaps half of the cases. However, most boards for commercial and industrial products aren't subject to the extensive level of testing that would be needed. Consider a few examples:
- A board with hundreds of ICs, each with a decoupling capacitor. All the capacitors are connected in parallel to the power and ground. In-circuit testing would show the sum of all the capacitors but not the individual values. With values ranging from nano-farads to tens of micro-farads, it would be impossible to know whether a 1nF capacitor was used instead of a 10 nF. The board would also work fine in functional testing. But it would probably fail for EMI, which is not a production test.
- A coupling capacitor of the wrong value on a microprocessor. This could cause a slight delay before the microprocessor turns on, but it would turn on and the board would pass a flying probe test. It may then exhibit erratic behavior in the field, such as race conditions in the timing or failure to turn at certain temperatures.
- A board with the wrong pull-up resistor in a circuit. The device may function properly, but draw too much current. This would most likely show up as end user complaints about short battery life.
How does the wrong part get onto a board?
Occasionally, the manufacturer labels reels with the wrong part number. More often a distributor makes an error when creating a barcode label for inventory control and tracking through the supply chain. Sometimes, we make an error when we enter the part number in receiving. Occasionally an SMT pick-and-place operator loads the wrong reel into the machine. All are human errors, either making typos or misreading a long number (often from a long list of numbers). All are very hard to avoid.
Ask any manufacturer when they last built a board with the wrong part, and the answer will probably be within the past than six months. Ask what procedures they have put in place to prevent it from happening again, and the answer will probably involve more human intervention: checking and double-checking the data entry and barcodes and machine setup.
This is not inherently a bad thing, but human intervention is usually the most expensive and least reliable solution to this sort of problem.
A better solution is to set up automated testing to verify the component value before placing them. At Z-AXIS we use pick-and-place machines from MyData that have the option to test passives for correct value — and for diodes, also the correct polarity. By enabling this option and checking several components every time a new reel is loaded, we avoid placing wrong parts without adding a lot of labor cost or significantly effecting throughput.