>My guess is that nobody packaged untested silicon, but
>rather that post-packaging testing wasn't done. I would
>expect that some small percentage of parts are messed
>up at the packaging stage, and perhaps this happens in
>a bursty manner (oops, there was a spot of dust on the
>lens of the robot vision system that was controlling the
>wire bonding...) and so a particular batch of a particular
>component could have problems.

Sounds reasonable. I guess one could also get a percentage where the bond
could be good, but the pressure of forcing the epoxy into the mould either
breaks a bond, or finds a weakness in the wire at some point other than the
bond, causing a break.

As to shipping packaged but untested parts, I guess that the bare chip
testing is not full parametric testing, but basic functional tests, and that
the full parametric tests occur at a plant with labour which has a higher
level of technical expertise (read more expensive) than the assembly plant,
so shipping assembled, but non-parametrically tested, chips may well be a
viable economic way of doing things.

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist