Abstract :
In the real world, no component manufacturer can attain loo-percent yield of final product (regardless of how well the device is designed, the scrutiny of incoming materials, or the care taken in assembly)-the user then must decide whether he can tolerate standard product quality levels. This will to some extent depend upon the supplier´s integrity, to a large extent upon the interest in reliability shown by the user, and by the complexity of the user´s system. The writer feels that for certain categories of components (resistors, capacitors, most low-current diodes, logic or switching transitors, etc.); screening, burn-in, and/or 100-percent testing is not economical. Other specific categories such as: IC´s (in reality subsystems), power devices SCR´s, Triac´s, transistors, etc.), hybrids, memories, microprocessors, etc., lend themselves to the philosophy of paying an insurance premium (screening, burn-in, testing) at the lowest level of assembly. Too often the true costs of board, system, or field repair is swept under the rug for personal or political reasons. This paper is intended to cut a swath through the statistical and accounting morass to enable some sound technical and manufacturing decisions to be reached.
