costs and few benefits. The truth lies somewhere in between these two extremes. The ideal solution is a sound economic model that lets designers-who have products with various production volumes, design complexities, specifications, support requirements, and customer stress levels-better determine what type of BIST, if any, is best for their applications.IC manufacturers evaluate the benefits and costs of wafer-, intellectual-property-, system-ona-chip-, and IC-level BISTs based on profit margin impact. Board manufacturers typically focus on the economics of boundary scan and boardlevel BIST implementations, although they should also consider the benefits derived from IC-level BIST. Board-level BIST, a relatively new phenomenon, 1 integrates IC-level, boundary scan, and board-level logic BIST for both circuit card vendors and system integrators. System manufacturers consider system-, board-, and IClevel BIST in determining the economics of the Economics of Built-in Self-Test International Test Conference 70 Using built-in self-test at the right level offers users significant cost savings, but determining which level, if any, is best for BIST can be complex. A detailed economic analysis can unravel heterogeneous costs and benefits so that designers and managers can make the right decision. Design and test Design, design for testability (DFT) and test time, budgeting considerations, marketing, quality and reliability, target throughput, training, packaging technology, circuit layout, silicon, pin out, CAD tools, designers' skills, performance degradation, computing equipment, and capital investment DFT and ATE Purchasing, depreciation and maintenance, operator and programming salaries, other labor rates, setup time, test program length, pin memory, program fault coverage, program preparation time, test and diagnosis time, test fixtures, manufacturing yield, production volume, and number of board types Field service Repair depot, logistics support, field repair, downtime, repeat visits, and design for maintainability