Issue link: https://iconnect007.uberflip.com/i/701250
54 The PCB Magazine • July 2016 Method At one or more defined measurement points the reference standard is measured at least 25 times under repeatable conditions (one opera- tor, normal production measurement condi- tions). Statistical software is generally used to ana- lyze the data. Capability indices are calculated: • The capability of the gauge is given by: C g = (K/100% * Tolerance) SV Where: K = percent of the tolerance for calculating C g , default = 20% Tolerance = USL - LSL SV = study variation given by the standard deviation (n – 1 formula) of the n measure- ments multiplied by 6 • The capability of the gauge, considering both the gauge accuracy and precision is given by: C gk = (K/200% * Tolerance) - |X – X m | SV/2 Where: K = percent of the tolerance for calculating C g , default = 20% Tolerance = USL - LSL X = mean of n measurements X m = reference measurement SV = study variation given by the standard deviation (n – 1 formula) of the n measure- ments multiplied by 6 The default value of K = 20% has been used for several decades. Any positive value can be used if it meets the requirement 0< K< 100. Us- ing values between 10% and 20% is common. The K = 20% is considered the "golden rule", and asserts that the measurement uncertainty (U mea ) should ideally amount to at the most one tenth of the tolerance [6] . Measurement uncer- tainty (U mea ) is the margin of error that that ex- ists about the results of the reference value. At the time this rule was formulated, it was considered that if it was adhered to the mea- surement uncertainty could be ignored. This is still widespread today. If measurement uncer- tainty is of specific concern with tighter specifi- cation tolerances, then values closer to K = 10% are recommended. The assumed "width" of the measurement system distribution is 6. Hence the default val- ue for study variation is 6σ, which contains ap- proximately 99.73% of the values. The Automo- tive Industry Action Group (AIAG) recommends the use of 6σ in gauge R&R studies [1] . Some com- panies use 5.15σ which contains approximately 99% of the values [7] . Using 5.15σ is conditional AABUS. Minimum acceptable capability indices [8] : Cg > 1.00 Cgk > 1.00 Preferred acceptable capability indices [2] : Cg ≥ 1.33 Cgk ≥ 1.33 Example NIST traceable standards of electroless nick- el immersion gold are used. Calibration uncer- tainty, gauge resolution, measurements, analy- sis, and capability indices are computed. Calibration Uncertainty: Standard: 180.00 micro-inches Uncertainty: 3.15 micro-inches Confidence: 95% Nickel tolerance: 118.10 micro-inches Significantly less than 10%: (3.15/118.10) = 2.7% Standard: 2.60 micro-inches Uncertainty: 0.05 micro-inches Confidence: 95% Gold tolerance: 2.37 micro-inches Significantly less than 10%: (0.05/2.37) = 2.1% Gauge Resolution Analysis: Resolution: 0.01 micro-inches Nickel tolerance: 118.10 micro-inches Less than or equal to 5%: (0.01/118.10) = 0.008% Resolution: 0.01 micro-inches Gold tolerance: 2.37 micro-inches Less than or equal to 5%: (0.01/2.37) = 0.42% USING THE TYPE 1 GAUGE STUDY TO ASSESS MEASUREMENT CAPABILITY