Is it time to rethink PCA test strategies in consumer electronics?

This paper challenges the stafiis ytm in resting .stra/egie.y of printed cif-ciiit hoard assemh1ie.s. A paradigm shi/ifi.om testing qnuliry into prodiicts to hidding y t ~ u l i ~ t in o prodticts is proposed. Automution und udvunced de.vign and munnfactoring techniyires provide the n1ethod.v /hut enable a change h 7 the wav we think ol~ont testing, ' Index Terms ATE, PCA Testing Quality Improvement, Test Strategy. 1. ~NTRODUCTION D EFECTS in electronic products occur for one of three reasons. These are due to material related problems, design problems, or due to process related damage. Testing strategies should be built around these possible causes and should be targeted at reducing overall costs of producing products. Current test strategies were appropriate when introduced. Changes brought about by automation, improvements in design automation, quality control and reliability should result in test strategies being rethought to help reduce overall manufacturing costs. 11. HISTORIC TEST PRINCIPLES When printed circuit boards were manually assembled, quality systems were designed lo inspect quality into products rather than ensure that quality was built inio products. The scenario was that products had components and materials inspected at goods inwards, and in an attempt to minimize repair costs testing was conducted at strategic places in the production process. This test philosophy, which is still in used today, is to test quality into products, rather than to verify functionality and to help align and calibrate products. Manut:dcturers of autometic test equipment (ATE) justify the expenditure of complex and expensive test syslems with early fault detection (i.e. tesling quality into a product). The wellknown logarithmic rule is used, where the cost of repairs increases by a factor often for every step in the manufacturing process during which the problem is not identified. It is accepted as a way of life that production yield will be less than perfect. Increased product complexity and the associated complex testing problems is the other reason that justifies the huge investments in test systems. When the manufacturing process was largely manual such lest strategies were appropriate. With the highly automated processes that are high repeatable, do test strategies devised for a predominantly manual production process still apply? 111. FACTORS AFFECTING PRODUCTION YIELD In determining a test strategy, one must consider what one wants to achieve with the test or tests. According to Werner " ... In order to design the oprimrim test urrungemenrs one [ I 1 miissf idenlib thefuults the ferf svstem i.s ~ i n g to./; .r..." The probability of a defect occurring is higher the more complex a Printed Circuit Board Assembly (PCA) is. Complexity can be defined using Oresjo's Complexity Index PI: C, = ((#C+#J)/IOO} *S*M*D (1) Where Ci Complexity Index #C ~ Number of Components #J Number of Solder Joints S ~ Board sides ( I Double Sided, 0.5 -Single Sided) M ~ Mix (I-High Mix, 0.5 for a low Mix) D Density (Joints / Square cd15.5) If the resulting Complexity Index is below 50, i t is a low complexity board. A result between 50 and 125 is medium density, while a result above 125 is high complexity. The sum of the number of components and the number of solder joints (#C+#J) represents the number of fault opportunities. Factors such as board densities (number of solder joints per cm2), type of soldering process, component size. board layers and component mix has an effect on the probability of failure. The volume of the production run is an important factor in fault prevention. The larger the volumes the lower the number of faults produced over time, as the process engineers have time to reduce process related problems. Figure I shows the results of a study on printed circuit assembly fault spectrum of over 85 000 inspected boards. ' Cenlml Queensland Universily, Sydney Inlcmational Cenlrc, Australia 0-7803-8526-8/04/$20.00 02004 IEEE 523 Fault Spectrum Inspection (AXI) that have evolved in addition to electrical tests of In Circuit Testing (ICT) and Functional Testing (FT). Current practice is to use a combination of these testers to obtain as high a yield as possible prior to FT or Systems Test. The test strategy employed is a mix of test systems mentioned before. Much has been written about what a test strategy is. As mentioned before, the problem is that our mentality is to test quality into products, rather than to build it into products. High volume production, as is generally the case in consumer electronics, allows for improvements in processes and it justifies the extra effort required by R&D. Production yield improvements start with component selection, R&D improvements, and process improvements. Automation is a driving force in these processes, as it permits greater eficiency and repeatability. Fig. 1. Graph 1 PCA assembly Fault Spectrum Consumer electronics is generally a high volume business, with high density boards. The high density is due to miniaturization, and the complexity of the boards is constantly increasing. Knowing the fault spectrum helps identify process points where corrective action can take place. It is well known and agreed upon, that the sooner a defect can be found, the lower the cost of repair. Knowing the fault spectrum can help reduce the defects induced during the manufacturing process. The issues of board complexity discussed thus far and their relation to defects is in agreement with current manufacturing philosophies. These current manufacturing philosophies have one shortcoming, they accept defects will be produced, and that to prevent defects from reaching the customer quality has to be tested into the product. It is time that our thinking about manufacturing changes, from testing quality into a product to building quality into a product. The American quality Guru Phil Crosby said that only zero defects are acceptable. If the consumer electronics industry accepts this, without adding any “buts” substantial improvements can be made in cost reduction, quality and reliability. So what do we need to do in order to make this paradigm change?