Issue link: https://iconnect007.uberflip.com/i/701250
52 The PCB Magazine • July 2016 study, the resolution of the gauge should be less than or equal to 5% of the tolerance. Reference Value The reference value is the known and cor- rect measurement associated with a standard or part. It serves as a master value for comparison during the Type 1 Gauge study. Reference val- ues can be determined in several ways depend- ing on industry standards and company and customer expectations. The preferred reference values are either: • Standards traceable to NIST or equivalent • Values agreed upon by the affected parties (AABUS) If no appropriate standard is available, a pro- duction part should be calibrated and marked as a standard. Preferred methods of establishing a reference value from a production part include: • Average of repeated measurements from a more accurate measuring gauge • Focused ion beam (FIB) analysis and measurements after the gauge study is completed In either case, the standard or production part used as the reference value must meet the following criteria: • Permit an unambiguous result of measurement • Be stable over a long period of time • Have the same characteristics as the parts to be subsequently measured with the gauge • Conform to qualitatively higher standards: – Calibration uncertainty (U cal ) must be significantly less than 10% of the difference between upper and lower specification limits, so that C gk ≥ 1.33 can be attained [1,2] . • U cal is the margin of error that exists about the results of the reference value. In colloquial terms, this would be expressed as "give or take." For example, an EN standard might be 150 micro-inches thick "give or take 3 micro-inches." • Two numbers are needed to quantify U cal [3] : – Interval—the "give or take" margin – Confidence—how sure we are that the 'true value' is within the margin – In the EN example above: 150 micro- inches, ± 3 micro-inches at a level of confidence of 95% • Reference value intervals need to be known, preferably at 95% confidence The reference value should be as close as possible to the center of the specification limit for the characteristic you are measuring. Ideally, the reference value should be within 10% of the center value of the specification limit. At a min- imum, the reference value needs to be within the specification limit. Tolerance Value Always consult current standards for current specifications. The specifications cited here are example values only. IPC-4552A Performance Specification for Electroless Nickel/Immersion Gold (ENIG) Plating for Printed Circuit Boards lists the following values for plated thicknesses: Nickel: Maximum: 236.2 micro-inches Minimum: 118.1 micro-inches Gold: Maximum: 3.94 micro-inches @ +4σ Minimum: 1.57 micro-inches @ -4σ Tolerance values are calculated as follows: Nickel: (236.2 - 118.1) = 118.1 micro-inches Gold: (3.94 - 1.57) = 2.37 micro-inches Opening up the specification limits would artificially inflate C g & C gk without ever improv- ing the gauge itself. Generally, it's a bad idea to let the process or the measurement instrument dictate the tolerance [4] . Nevertheless, increas- ing the tolerance range and keeping the study variation (K) of the gauge fixed, does imply that measurement uncertainty takes a smaller chunk of the tolerance range, which is in fact indica- tive of better performance of the gauge [5] . USING THE TYPE 1 GAUGE STUDY TO ASSESS MEASUREMENT CAPABILITY