Printed Circuit Board Assembly & PCB Design SMT Electronics Assembly Manufacturing Forum

Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.


here's one for you board stuffers!!!

Tim

#13081

here's one for you board stuffers!!! | 6 January, 1999

Any idea what the "industry standard" is for fall out on assembled PCB's? I realize that the controls put in place in your processes will effect your overall quality. But I'm trying for a general answer. On pre-tested assemblies, should you expect to find defects in 5, 10 20% or your assemblies? Ideally you would like 0%, but no one is building those ( if we were, we wouldn't need to test them.) What's a common percent? If I find that 10 of out 100 tested need minor rework, is that high or about right? Any input you could offer would be appreciated. Thanks for your input.

reply »

Steve Joy

#13082

Re: here's one for you board stuffers!!! | 9 January, 1999

| Any idea what the "industry standard" is for fall out on assembled PCB's? I realize that the controls put in place in your processes will effect your overall quality. But I'm trying for a general answer. On pre-tested assemblies, should you expect to find defects in 5, 10 20% or your assemblies? Ideally you would like 0%, but no one is building those ( if we were, we wouldn't need to test them.) What's a common percent? If I find that 10 of out 100 tested need minor rework, is that high or about right? | Any input you could offer would be appreciated. | Thanks for your input.

That's a tough one. First, no matter what it is, it's never good enough! haha Next, realistically, it is going to depend on a total cost model. If you are placing CPUs and can't rework them (well, we can get Steve Gregory as a consultant) them the board cost is minimal compared to the potential assembly scrap cost. In this case, we might ask for proof of 1500 DPM incoming for all visual/electrical/mechhanical defects. Since board manufacturing is to some degree a sorting operation, you would have to add overhead to insure the quality. It is really a difficult problem if you think about it. Buyers are always asking for cheaper, better boards which naturally drives overhaed at the board shop. The buyer may go shopping to get the price cost down, but that may effect the quality, so the supplier is driven to comply, which adds overhead, and on and on.... Anyway, the short answer is that the AOQL really depends on the cost of quality. This is why it is so important that the specs accurately reflect what is good and what is bad. The spec is always negotiable for visual defects and if the PCB shop has statistical proof of equivalency, I will change the spec. Hope this helps, Steve

reply »

Russ Steiner

#13083

Re: here's one for you board stuffers!!! | 12 January, 1999

| Any idea what the "industry standard" is for fall out on assembled PCB's? I realize that the controls put in place in your processes will effect your overall quality. But I'm trying for a general answer. On pre-tested assemblies, should you expect to find defects in 5, 10 20% or your assemblies? Ideally you would like 0%, but no one is building those ( if we were, we wouldn't need to test them.) What's a common percent? If I find that 10 of out 100 tested need minor rework, is that high or about right? | Any input you could offer would be appreciated. | Thanks for your input. | Perhaps the best way to answer the question is two fold - First, the only real measure needs to be based on equal assemblies. For example, if you are building a leaded heat tape control (fairly simple) you may deal with wave soldering 12 x 18 inch single sided panels with 1/2W leaded resistors. On the other hand, when I'm building high density motor controllers that have 15-mil pitch, 0402's and the like, it stands to reason that I would have diffeent sets of solder defects to consider. Secondly, and more to the point (I think), your measurement in your question should compare the number of solder joints to the number of failures - such as parts per million (ppm) of failure. Then, compare your ppm numbers to those of others success running like component mixes and densities. Long answer to a simple question, but you really need an apples to apples comparison or your numbers won't mean much. Depending on density and components, 10% might not be bad - or, it might stink! Good luck! ;-) (Russ @ Controls/inc.)

reply »

Manufacturing Software