Toefuzz
1st December 2005, 05:58 PM
Okie doke... I am working with various millage gages (test coating thickness) and was wondering how accurate my data actually was. The gage has approximately a 10% margin of error built in. After performing a gage r and r my results were something like 9.8%, 20%, and 45% with various gages. Obviously the 9.8% is the preferred gage, but what does this actually mean? If my coating is 2 mils thick does this mean my reading can be off by 50% (nice round number for easy math? If so, does that mean my millage could have actually been 1 to 3 mils? Sorry if these questions seem silly but I'm trying to be prepared in case my customer asks the right (or wrong depending on your viewpoint :)) questions.
Miner
1st December 2005, 08:58 PM
Okie doke... I am working with various millage gages (test coating thickness) and was wondering how accurate my data actually was. The gage has approximately a 10% margin of error built in. After performing a gage r and r my results were something like 9.8%, 20%, and 45% with various gages. Obviously the 9.8% is the preferred gage, but what does this actually mean? If my coating is 2 mils thick does this mean my reading can be off by 50% (nice round number for easy math? If so, does that mean my millage could have actually been 1 to 3 mils? Sorry if these questions seem silly but I'm trying to be prepared in case my customer asks the right (or wrong depending on your viewpoint :)) questions.
You have not really provided enough information. What is the 10% margin of error? Is that the manufacturer's stated accuracy? That is, +/- 10% of the scale reading?
Regarding the 9.8%, 20% and 45%, are these P/T Ratio (%Precision to Tolerance), %GRR, or % Contribution?
P/T ratio is the percent of tolerance taken by the measurement error. < 10% is excellent, <30% is acceptable for part inspection purposes.
%GRR is the percent of product variation taken by measurement error. < 10% is excellent and < 30% is acceptable for process control (e.g., SPC) purposes.
% Contribution shows how much each component of variation contributes to the total.
See these threads for additional detail.
http://elsmar.com/Forums/showthread.php?t=13794
http://elsmar.com/Forums/showthread.php?p=125257#post125257