Blog Layout

Are Six Sigma calculations incorrect?

Husain Shekhani • Jun 26, 2021

If you use statistical definitions of the standard deviation to calculate variability, you will readily calculate that +/- 2 sigma (standard deviations) would result in ~5% defective product. +/-2 sigma  encompasses ~95% of all the samples. 5% defective product for transducer is not bad, as long as you can catch 100% during manufacturing testing. This is definitely normal for many ultrasonic transducers assemblies.

That being said, what you will usually read in "Six Sigma" training materials that a two sigma process results in 30% defective product. So what is going on? Why is there a difference?

Six Sigma methodology adds an extra 1.5 sigma to account for variation in the process in the long term. Obviously, we cannot do experiments and tests today that will tell us the variation of tomorrow. Therefore, we assume 1.5 sigma to be the deviation in the mean of the process in the long term. The actual amount may be much more, or in a rare event the variation will be less.

Don't assume that the transducer testing that you have just completed in the lab, even if it has 100 samples, will properly describe the long term results. There are so many sources of deviation, which include test methods, crystal property variation, and variation in actual part tolerances (both dimension and features).

Follow the link here to learn about my ultrasonic transducer consulting services and prices

Services and Prices

Did you like this post?

Sign up for my email list and receive my free guide to the:

 "Top three reoccurring challenges in ultrasonic transducer development"

    Share by: