I have a question about inspection, touch-up, and how far you need to go to get a product that meets the required IPC-spec. According to some reports (i.e. "New Study Reveals Component Defect Levels" by Stig Oresjo of Agilent) the average defect level in PCB assembly is around 1100 dpmo. Even with much better dpmo-figures it is easy to see that except for very simple boards almost every one of them will have one or more defects before touch-up. A significant portion of these defects will be visual defects like an isolated spot of flux residue, overhang of an occasional bent lead, micro-solderballs, an occasional insufficient solder, an overlooked spike from handsoldering, etcetera. Although these are all real defects according to IPC-A-610, the chance that they will cause the product to fail during their service-life is minimal.
Now the point of this posting is: how much effort do you need to put into finding and correcting visual defects. You could argue that every board that still has a visual defect simply does not meet the IPC-requirement, and therefore every boards needs to be 100% inspected and reinspected until you can be sure that you have found and reworked all visual defects. The result is a nice-looking product which meets all the inspection requirements of IPC-A-610, but unfortunately it will be an expensive board. Another approach is to visually inspect the boards mainly for the purpose of feedback to the assembly process (on sample basis), and rely on electrical tests for finding defects to be reworked. Quick and effective, but only some of the visual defects will be found and thus, formally, the product (or at least some of them) will not meet the inspection requirements of IPC-A-610.
I try to convince others that we spend much more time on inspecting than we should, but so far every visual defect that is found is regarded as proof that I am wrong. I am interested to hear how others are coping with this issue.