An often overlooked discipline for component engineers is the retrieval of failed components from customer service technicians. Many times when products are sent back for repair, the technicians fail to keep running records of the failed components. Or, worse yet, they fail to push the trouble-causing parts back to the engineers for evaluation.
Valuable information is lost when the same components keep failing and no one initiates further investigation. The technicians unsolder the bad parts and, more often than not, toss them into the waste to keep their work surfaces clean.
In the past, I have alerted customer service to keep any failed components in a separate box and to keep a record of how the component failed and under what conditions. When I examined the box and saw more than one component of the same part number, I would ask the repair technicians if the failure symptoms were the same and how often they had returns for the same problem. If the parts had the identical date code and there were no failures from other date codes, I would call the manufacturer to see if other companies had reported similar issues involving parts with similar date codes. Many times, the manufacturer would ask for the parts back, so it could evaluate them itself. This would not be possible if we had not kept the parts.
This best-practice continues with the completion of a failure mode analysis request form. When the manufacturer receives the form and the failed parts and acknowledges the receipt of both, the official failure mode effects analysis (FMEA) begins. The component engineer is responsible for tracking the progress of the off-site analysis and informing the key players as information becomes available. Meanwhile, the design engineers are told of the suspect part, so they can decide whether to use components with that date code in prototypes. The materials department is also advised and given the option of purging its work in process inventories.
Meanwhile, the assembled boards where the parts are used are considered suspect. If it is possible and practical to do so, different date codes are substituted for the suspect parts. If it seems like a lot of trouble, consider the potential trouble and costs that are being avoided. At the very least, if no action is taken other than beginning the FMEA effort, at least there is a record that an attempt was made to flag what could be a very dangerous situation.
If a component's failure has nothing to do with the lot number or date code, look for modifications made after engineering release that may have changed the stress factors. A slight increase in voltage or current due to a substitute part used elsewhere on the circuit may be pushing your component beyond its specified operating limits.
In reality, there could be a dozen causes for the failure, but when a device with a single component part number seems to be failing over and over again, there is a common cause that should be identified as soon as possible. Having customer service keep the errant parts is the only way to make the failure analysis study credible.
If your company is not doing this, consider implementing this simple direction. This will allow you not only to catch the failure trends now, but also to avoid repeat performances in the future. Customer service will appreciate that its reports are valuable resources that will reduce failure rates, returns, and unhappy customers.