Testing Method

[aathi1]

Hi All,

I'm working DOE project on leak test and tranfer the product to factory. We facing high leak failure due to design issue. Meanwhile waiting for new design change , the management want to increase the spec so that we have alot of passes. My question is as follows:-

Is there any statistical method that I can use instead of increase the spec.
I want to test the product using original spec vs increase spec. Is that method can tell us something?

Balan.G

 

[melone]

Before condemning management for "changing the spec", do you understand the details of why the spec was set where it was? Obviously, changing the spec just to arbitrarily inflate the passing results will undermind the entire purpose of measuring the value. However, if the current spec limit has no emperical basis, then perhaps they are simply adjusting the limits to remove unnecassry part rejection (alpha rejection). Here is an example:

I am a pencil manufacturer, and I need to make pencils that are 8 inches in length. One of my final inpsection items is to measure the overall length of the product. However, I defined the length to be 8.000 +/- 0.001" . This implies that I need a piece of measurming equipment that has at least 0.0001" precision. That means I need to use a Coordinate Measurement Machine (CMM) to measure the LENGTH OF A PENCIL. Doesn't this seem like overkill? Perhaps, it makes more sense to define the length to be 8.0 +/- 0.1". This will allow me to use a digital caliper to measure the value. Furthermore, there is no real benefit to have a pencil that is guarenteed to be 8.000" when a pencil that is 8.1" will function just as well.

I don't know the specifics of your scenario, however, is it possible that they are simply redefining the length of the pencil to something more realistic? Do the current test limits guarentees that functionality of the end product is met?