Issue link: https://iconnect007.uberflip.com/i/340751
72 The PCB Magazine • July 2014 by Rick Kaim ally SuPPly inc Isolation Testing and Adjacency Whether you have a cursory un- derstanding of adjacency or the subject is new to you, this article will explain the past and current state of adjacency testing and where it is heading. Why adjacency means dif- ferent things to different people is simple; until recently it had not been adequately defined by an industry standard body. IPC- 9252 (Febru- ary 2001) section 3.1, Adjacent Features, is not an explanation of adjacency used by moving probe technology. The next revision of 9252A did address adjacency and set an institutional minimum standard; important yes, but that does not guarantee a full understanding by the industry, and most importantly, users. Until recently, moving probe manufactur- ers have set their own standards and more im- portantly CAM software providers have devised their own best methods. Adjacency distance value (ADV) is the distance from an edge of any feature on one net to an edge of any feature on another net. The proper ADV is important to many stake holders, from prober equipment manufactur- ers and the CAM software suppliers, to the board shops, their customers and finally, the end-users. None of these entities are myopic; for instance, a board shop has both a quality manager and a production manager, but these are opposing forces in the world of adjacency. What is the best compromise for end-users and board manufactures? Let's look at adjacency's past, present, and likely, future. What adjacency has done in its past is truly remarkable; it has allowed flying probes testers to enter the mainstream of bare board electrical test. By using logic to overcome a physical limitation, a new industry within an industry was born. The other major influence in mainstreaming probers was signature (capac- itance) testing, which is a separate subject from adjacency. The basic need was to tackle isolation itera- tion explosion. If each net was tested against all others the number of tests is [N x (N-1)]/2. So, for example (Figure 1). It is easy to see that a small number of nets can be tested completely, but a larger number of nets become unrealistic; even if a prober could test consistently at 10K test points per minute (and none can) the isolation test on the 10K net board would take more than 80 hours! Probers on average can test isolation in the hundreds of nets per minute, not thousands, in resistance mode because of the X, Y and Z motion need before the test can occur. The X and Y motion a r t i c l e