Flux Selection

Designed experiments I guess

reply »

Design Of Experiments (DOE). A statistical technique for determining the relationship between and relative importance of various factors controlling a process. So, it is a tool to help do things faster, better, and cheaper.

The two variants of DOE are: * RA Fisher is the godfather of classic array techniques * G Taguchi popularized Taguchi array techniques

Look in a recent �SMTnet Newletter� for a book review of �Six Sigma For Electronics Design & Manufacture� by Dr. S Shina. It covers a broad discussion of tools such as DOE that could help you improve your process. [If people are interested in this topic, contact Brian ( bpdoyle@smtnet.com ) at SMTnet and harass him into getting Dr. Shina to host an �On-Board Forum�. Just don�t let on that I suggested that you to bother him. He�s in a deep funk since the wheels fell off the Patriots� season and they have begun to look like the Lions and Bengals.]

Another very good book on the topic is: �Reliability Improvement With Design Of Experiments� Condra, LW; Marcel Dekker; 2nd e; 2001; ISBN 0824705270. This book is more focused than the Shina book. It is aimed at developing an understanding of the uses and potential applications of DOE. It is not aimed at teaching you to be a DOE wizard. Condra does an excellent job of producing a �readable� book. Many similar books are mind numbingly difficult to understand.

Stat-Ease [ http://www.statease.com ] sells very sophisticated DOE software and excellent training. Every so often, they send us a newsletter. It is one half self-promotional and on half educational. The educational side usually involves using their software to solve some earth-shattering dilemma. Such memorable dilemmas have included: * Sixth Graders Experiment with �Frisbee�s� to determine the effect of open ring / closed type, color, and throwing style on accuracy and distance. * Popcorn Shootout to determine optimal microwave popcorn factors * Lutefisk, ymmm, recipe factors. [Although I remember this one fondly, I don�t find it on their site. So, it could be that I�ve got a hankering for that delicacy and am wishing they�d run a DOE on it. Whatever!!] [More worrisome, �Google� produces no link to stat-ease AND lutefisk (or the Americanized 'lutefish').]

By pimping �Stat-Ease�, we: * Pose it simply as an alternate source of information. * Receive no benefit from Stat-Ease, nor its principal, nor its subsidiaries. * Don�t mean to imply: You need �Stat-Ease� to do DOE. [We've used Excel and a free demo disk �WinRobust Lite�( http://www.winrobust.com/ )]

reply »

Thanks guys, now I know what DOE is... I learned something new today :) Patriots will come out of it and get into the playoffs. yuk, yuk, yuk.

reply »

MA/NY DDave

You say, "I won't list the other publishers yet DaveF did list one of the good publishers."

Why won't you list the other publishers?

reply »