Test: when is enough enough?

The focus of this panel is the trade-off between excessive testing and acceptable shipped quality levels. How much testing is enough? We all strive to ship our customers high quality product, or at least product that meets our customers’ quality expectations. Often improvements in shipped quality levels can be made by plugging obvious holes in a device’s test suite, sometimes simply by increasing the fault coverage of an existing test. However, at other times, there are no obvious deficiencies in the test methodology and the hunt begins for new failure mechanisms, fault models, and test methods to detect them. Many times the enhanced tests not only improve the shipped quality level, but also drive an unacceptable decrease in yield. This decrease in yield is the result of catching defects that would have appeared in the field (the intent of enhancing the manufacturing test), but also can be the result of an over zealous culling of manufactured product which results in scrapping product that would never have failed in the field.