Stability of process is what your measuring so lets not assume a stable process(if your process was perfect you would have no defects). With no test a medium complexity board will be around 68.7% (according to Consulogic). I dont have many other sources and they very dramatically depending on how they measure potential defects. Basically it goes up with complexity and opportunities for defect.
i.e. if you have one 144 pin QFP you have 144 potential defects plus pin one orientation and one opportunity for missing part.If you place seventy two 1206 resistors you have the same solder joint count but 72 possibilities for missing part. If you cant place and solder 1206 resistors your in trouble,yet the statistics are with you, if you get an ocasional bridge or bent lead on a QFP your probably average. If your running bottom side surface mount through a wave, unless you have great design guys, expect a few problems.
It would be great if there was a weighted scale on how to grade defect opportunities by actual difficulty but im not aware of any. Most people use PPM so numbers look good, we use FTY to understand what customers would see if we did not test or inspect but that makes it hard to measure the statistical impact of process improvements. We are working to look at both.
IPC has IPC-9261 and IPC-7912 check them out.
I would love to hear others comments. Most likely no one will share numbers out of fear of customer misinterpretation. We do a number of complex boards with over 10% first pass failures and the customers are always freaked out.
reply »