Jan,

The problem is not with destruction of parts on the board. It's with the fact that when you solder a high-pin-count part on the board, one with a fine pitch, you not only waste that part if it doesn't function, but it causes waste of the other parts, e.g. the $2000 FPGA, that are also soldered onto the board.

Passives are not likely to be different from one end of the tape to the other, so test a few at one end and you probably have a picture of what they've shipped you. If they're mislabelled, it will show up immediately.

Distributors are prone to ship you another customer's rejects, though. It's easy and nobody will notice if you don't inspect them right away.

I have examples of boards not a half-meter square that have 2500 passives and 400 or so actives on them, parts being placed on both sides of the board. The one I'm thinking of was very costly, with several multi-hundred-dollar XILINX FPGA's on it and a couple of pretty expensive microprocessors, too, not to mention considerable memory, which was very costly back then, and most of that hardware was soldered in place rather than socketed. Back then, one could desolder a part and replace it without much ado, though it did cost half a day's work for an experienced technician. If you have CS/BGA packaged parts, though, and most of the really low-priced ones are, you don't have access to all the pins. You can remove the, say, 1148-ball FPGA, but then have to have it reballed in order to reuse it, and, of course, it's not alone ... there may be other parts that cost quite a bit too. Some of the boards I've seen as FPGA demo boards cost $50000. If there's a non-functional MCU, with critical functions on which the circuit relies, over in the corner, it makes that entire board useless. That's the problem I'm trying to avoid. When circuits cost that much in units, you don't want to have to repair them. First, of course, because it is costly to rework a high-density board, and secondly because it's damaging to your reputation if your $50K board doesn't work where your customer can see it. If you're smart, of course, there's thorough OUTGOING inspection, too, so at least your customers don't see the boards that don't work. However, you'll probably invest man-days in isolating the fault and even more in repairing it. Incoming inspection, particularly if it's random, is much less costly, as it can be planned and will have little impact on schedule and budget.

Incoming inspection of simple passives and linears isn't a problem, but as the complexity of digital components increases, it's important to know BEFORE THEY'RE SOLDERED IN PLACE, that they're fully functional.

RE