the reason you don't want to use PPM is exactly the reason you need to. You cannot compare a square 1" block to a multi faceted custom machine part. In SMT assembly you will usually find that all 0603 and larger discreetes,25mil pitch, etc. are probably close to 0 defects while 0201, 16mil pitch will exhibit a higher defect level. You need to ask yourself what do I want to get out of this data? prediction of yield? etc... For example, If I have a board that has only 1206 and 50 mil pitch with a part count of 200 or so I won't even have it inspected before reflow. If i have a board that is double sided with multiple technologies and 5000 parts on it I will need to look at it a little more closely. This all figures out based on the suppliers Process capability. If this supplier on 1206 resistors has a defect rate of 100 DPMO they can expect 100 defects per million placements. If they have a board with a million placements there will be 100 defects on that board, 0% yield whereas a board with 1 placement will have a yield of 99.99+ %
This is only the only "fair" way to rate suppliers in PCBA manufacturing as far as I know. We once used to be graded against a plastic molding shop that made the box that our board went in. This was a square box with 4 screw holes in it. our PCBA was double sided, multiple component types, elastomer pads that had to be flat, 50 wires soldered on, glue applied, etc... Who between us had the better yield at the customer? Obviously the plastic guy. Once DPMO was established the quality levels between us and the plastic guy was much closer 50 DPMO vs 85 DPMO as opposed to 99.9% yield vs. 95-97%. another thing to look into is the design of your products to ensure that they are 100% manufacturable and "special steps" are not needed to produce this product.
Hope this helps.
Russ
reply »
All � I am trying to change my rating method for end item as...
- Jan 28, 2004
by
danielinsa
the reason you don't want to use PPM is exactly the reason y...
- Jan 28, 2004
by
RDR
