AQL values....how low can you go ?

Joe: Lemme see, you�re talking about controlling your manufacturing process within 4.6 defects per 100 opportunities or a 4,600 dpmo rate, a cumulative 99.95 acceptance rate, or a �2 sigma process, right? So, it�s fairly basic stuff, eh?

First, you say "I would like to know what AQL's are being used in the different markets, (military, telecommunucations etc.)." Implying that different markets have varying tolerances for poor quality products, which maybe true, but I think (1) you�re approaching this backwards and (2) this is probably not the correct audience to get meaningful answers to such a question.

On point (1): Your marketing people should be saying things like � "for us to sell our widgets into the blank market, 80% of sales in the that market require minimum quality levels of bla and for us to take market share, we need a product with bla+1 quality. Team, can we do it?" Then, the production guys and girls should assess their process and determine if they can or what it will take to do bla+1.

On point (2): General Electric and Motorola talk 6 sigma. There�s a couple of poles for your tent.

Continuing, you say "What kind of parameters (process control) are needed in the production line, (from SMT to wave soldering), to achieve a AQL of 0.25?" Every process has different capabilities and every machine within those process has varying capabilities according to machine age, hours use, maintenance, and a host of factors. You�re goofy to think that anyone could possibly answer the question. But anyhow, you need in-control and capable processes to achieve a AQL of 0.25.

More specific answers are:

1 We use AQL at incoming inspection with a sampling level selected based in supplier performance. Good guys and girls get sampled less, bad guys and girls get sampled more. We start with a C = 0 plan for a new supplier. 2 We don�t use product sampling as a direct way to manage our process, because information about whether our process is "in-control and capable" is too far removed from the inspection data.

� You know, it�s tough to tell, when looking at a SMT solder bridge, if the bridge is from goopy paste, mis-registered print, dirty stencil, poor component positioning, spastic handling, long wait time, high air flow in the oven, er what!!! � If you just look at AQL inspection results, it tough to pick-out where the process is trending. � If you just look at AQL inspection results, it�s very difficult to tell if the process is in control.

Without those things, the problem is huge and with people tending to work in compartments, it just becomes tough to drill-down to the source of the problem. And nothing gets done.

We monitor and chart important process parameters and take steps to keep them within process limits.

My2�

Dave F

reply »