Posts: 2
From:Bangalore, India
Registered: Jun 2001

posted 29 June 2001 05:47 AM               

---

We are analyzing the measurement system at a facility that manufactures engine valves. Data collection for linearity and bias studies have been done in accordance with the MSA manual. What inferences can be arrived at from the attained values of slopes and bias? Are there acceptance criteria/standards on which conclusions can be drawn

IP: Logged

Posts: 10
From:Welland, Ontario, Canada
Registered: Mar 2001

posted 06 July 2001 11:33 AM               

---

I posted a query on acceptance criteria for linearity in May in this forum. I have heard of two reasonable sounding criteria. One is linearity as expressed as percentage of the part tolerance range should be less than 10% (sort of an extension of the old 1/10 resolution rule of thumb). The other criteria is, measurement uncertainity in the reference standard (usually a very small number) + Gauge R&R + linearity or bias must be less than 30%. Using linearity if the measurement device is used over a range or bias if it is used only at one point in the range. All of these above are expresed as a percentage of the tighest tolerance range on the part/drawing that device measures. None of these is official but as long as you document these criteria you will in all probability be one step ahead of the auditor. If you think about it, this means that 30% of the tolerance range is being consumed by measurement uncertainity alone! I hope this helps.

IP: Logged

All times are Eastern Standard Time (USA)