| ??? 01/17/07 19:59 Modified: 01/17/07 20:01 Read: times |
#130998 - let me reframe the question ... Responding to: ???'s previous message |
Jan Waclawek said:
Richard Erlacher said:
How do you determine whether they are acceptable? Of course this depends on the application. What??? Do you mean that a part with internal malfunctions are OK so long as, initially, they don't impact your application? There is no universal answer as there are infinite potential modes of failure of each of the zillions of parts used in electronics today. It's not the zillions of parts on the market that I asked about, though. It's just the 805x system. I think it's quite appropriate for certain type of products to simply perform the smoke test, then configure (burn) and perform the "try to wiggle with all buttons" test. I can't think of one case, aside, perhaps, from a "one-off" device, where that would be adequate. "Smoke test" is for devices that were built on the kitchen table and are going no further. Real testing involves thinking about failure modes and stressing the system under test, within prespecified limits. If it's a board with "$2500's-worth of other components", in 100s series, there might be a JTAG-enabled (read boundary scan enabled, I know it's not the same nowadays, unfortunately) part and you might have developed some sort of testing via the JTAG. You need to perform the smoke test first, anyway; but this might save some hassle with shorted buses and similar stuff. I'm including in my question, all the boards that are being built 1000 per day, for the past ten years, into which established components are being used, and not "new" systems never-before-tested. I'd not suppose such a bad part which would burn out other parts (except voltage regulators and alike, but that would be too bad if those would not be reliable enough, anyway - I'd thrash the supplier or manufacturer of such). Unless you actually ARE producing 1000 units per day, the supplier and manufacturer don't care at all whether 10% of your units don't work because they've shipped you trash. All this might be completely inappropriate for your needs and application though.
JW PS. I've written a memory test when a "sclerotic" batch of SRAMs slipped through production to the field, but I knew exactly the mode how they failed from the examined specimen so the test was specific. One thing I'm curious about is how people determine whether their ATMEL or other mfg's parts actually work as specified. My own experience indicates that, at least with ATMEL, there's room for doubt. RE |
