FALSE CALL REDUCTION

What is the total volume of parts processed?

How well are your models defined, are the BOM's allowing a wide range of components? Are you using a global component database or product specific?

reply »

The basics that I utilize are: 1. Add multiple images (as you've said) 2. Adjust the thresholds. If the standard threshold for a package in your library is 50% and you get a few parts where they measure 43% and 46% and 45% AND THEY STILL LOOK GOOD, then maybe lower that threshold to 45%. IPC says a part can be half off the pads and still be Class 2. So maybe ease up on your position criteria. I'm not saying to allow 50% off the pad, but 15% won't hurt.

reply »

I used an early Mirtec about 20 years ago. I think none of us had previous experience with AOI at the time. We of course had a learning curve abut Mirtec was great with support. But we had trouble with false fails and part of the problem was bad image saves. We tried weeding them out but it wasn't easy by that time. I totally nuked the library and started fresh with a set of guidelines for adding pass images. We of course had to build the library back up which was a lot of work but well worth it.

side note one time it failed a capacitor and the operator said that it couldn't tell the difference and passed it. The machine was able to tell the difference in color between 10n and 100n capacitors. He should have checked the reel not just passed the fail.

reply »

You don't say what inspection type generates the most false failures. Each type has its own remedy. We see a lot of solder false failures on components near the edge of the inspection window. Creating a new frame for that component often clears the issue. This slows the machine down but is faster than having to go back and review the false failure. We also see false failures when checking the marking on a component. Component manufacturers don't seem to care about consistency of the marking. Different fonts often appear on the same part when those parts are made at multiple locations.

reply »