1 PURPOSE
To establish a process for when and how Gage R&R studies on measurement systems are performed.
2 SCOPE
Measurement systems used in the manufacture of brush and brushless motors and electronic controllers including power supplies.
3 REFERENCE DOCUMENTS
QS-9000 Measurement Systems Analysis (AIAG)
GRR [Link] – Gage Study Information
4 DEFINITIONS
Gage – any device used to obtain measurements
Measurement system – the complete process used to obtain measurements
R&R – Repeatability and Reproducibility
Accuracy – Accuracy is the difference between the observed average obtained with a measuring system and the actual true average.
Repeatability – Repeatability is the variation obtained when the same operator uses the same gage measuring the same part
Reproducibility – Reproducibility is the variation obtained between the average of measurements by different operators using the same gage to measure the same set of parts.
Observer – Person responsible for organizing the study. The observer records the data and usually performs the analysis. The observer does not actually take any measurements.
Operator – Persons responsible for taking the measurements during the study. The operators chosen should be selected from those who normally operate the gage.
5 STATEMENT OF WORK
The purpose of gage evaluations is to assure that the measurement system being employed is adequate for the purpose it is being used. Gage evaluations are usually performed after Control Plans are documented and before Capability Studies are done. GR&R
should be performed in the following cases:
· when new equipment is evaluated/brought in house
· a change is made to the measurement process
· the gage measures a critical dimension
· the gage measures a characteristic for final acceptance of product
· there has been a change to the production process
· when problem solving efforts point to GR&R as a potential source of variation
Guidelines for selecting sample parts:
· Sample parts used in GR&R studies must be selected from the process and represent its entire operating range
· Manufactured parts should be selected randomly from the manufacturing process over a period of several days of production. No two parts should have been manufactured within the same hour.
· Purchased parts should be selected randomly from multiple boxes from at least two shipments.
· For short study and long study, sample parts must be within specification.
· For attribute study, 50% of the parts selected should be out of specification (i.e. No Go). Selecting No Go parts that are grossly out of spec defeat the purpose of the attribute gage study. The best parts to use for attribute gage study are parts that are
marginally in or out of specification.
Guidelines for gage selection:
· The gage used in the study must have a calibration that is up to date.
· The gage must measure to increments of full 1 significant digit greater than the specification. For example: if the spec is 1.3 – 1.7 mm, the gage must be able to read to the nearest 0.01 mm.
· Each operator must use the same gage to perform the measurements. For example: if there are multiple calipers used in the area, each operator should use the same one for the purpose of this study.
· Prior to beginning the study, all operators and the observer should review together the SOP for using the gage to measure the characteristic addressed in the study. If an SOP does not exist, one should be documented prior to the study. The SOP should
clarify how to measure, where to measure, the number of measurements, and whether the final result is an average, max, or min of the measurements. The group should measure a part together to verify all are using the gage in the same manner.
5.1 Long Study The long study should be conducted with 3 operators, 10 parts, and 3 measurements per part for a total of 90 measurements. The parts should be measured in random order each time. Since each part is measured more than once by the same operator,
repeatability can be isolated from reproducibility.
5.1.1 Select sample Observer – select 10 sample parts, number each, and cover the number with masking tape; Note: the operators should not be able to see the part numbers during measurement
parts
5.1.2 Arrange parts Observer – arrange the 10 parts in the following order:
Operator A, trial 1 4 8 1 10 3 6 7 2 5 9
5.1.3 Measurement Operator A – measure each part one time
Observer – record the measurements.
5.1.4 Continue Repeat the above two steps until Operators A, B, and C have measured each part 3 times, arranging the 10 parts in the following order:
Measurement
Operator A, trial 2 9 7 3 5 1 10 6 8 4 2
Operator A, trial 3 1 3 6 10 9 7 5 8 4 2
Operator B, trial 1 2 5 6 3 10 1 7 9 4 8
Operator B, trial 2 9 4 7 8 10 5 1 2 3 6
Operator B, trial 3 10 4 1 9 3 6 2 8 7 5
Operator C, trial 1 5 6 9 7 8 3 10 4 2 1
Operator C, trial 2 7 3 10 8 9 5 2 4 1 6
Operator C, trial 3 7 5 10 3 9 2 6 8 1 4
5.1.5 Data analysis Observer – Enter all data into a copy of the Long Form. Note: If the study conducted must deviate from this QSP, document the changes to the process in row 2 of the Excel Spreadsheet.
5.1.6 Judgement Guidelines for judging acceptance of the GR&R error:
· Less than 10% error, measurement system is acceptable
· 10% - 30% error, measurement system may be acceptable based on significance of attribute being measured, importance of application, cost of gage, etc.
· Greater than 30% error, measurement system must be improved
5.2 Attribute Study The attribute study is a quick way of determining the acceptability of go/no-go gages. It is usually conducted with two operators and 20 parts, with each operator measuring each part twice, randomly.
5.2.1 Select sample Observer – select 20 sample parts (it is desireable to have some parts slightly above and below both specification limits), number each part, and cover the number with masking tape; Note: the operators should not be able to see the part numbers during
parts measurement
5.2.2 Arrange parts Observer – arrange the 20 parts in the following order:
Operator A, trial 1 3 20 10 14 15 1 17 12 7 9
18 4 5 6 8 16 2 11 13 19
5.2.3 Measurement Operator A – check each part with the go/no-go gage
Observer – record G or NG.
5.2.4 Continue Repeat the above two steps until Operators A and B have checked each part 2 times, arranging the 20 parts in the following order:
Measurement
Operator A, trial 2 12 13 11 7 20 5 8 19 10 14
17 18 4 1 16 15 2 9 3 6
Operator B, trial 1 13 10 12 20 15 11 8 1 3 2
16 6 18 19 7 14 4 17 5 9
Operator B, trial 2 20 18 13 5 10 9 3 16 7 8
4 19 2 15 11 1 17 14 6 12
5.2.5 Data analysis Observer – Enter all data into a copy of the Attribute Gage Study. Note: If the study conducted must deviate from this QSP, document the changes to the process in row 2 of the Excel Spreadsheet.
5.2.6 Judgement The gage is acceptable only if all measurement decisions (four per part) agree for all 20 parts
Gauge R&R Long Form Study
Study Number: Part No.: Performed By:
Gauge Name: CMM Part Name: Operator A:
Gauge No.: Operation No.: Operator B:
Graduations: Characteristic: location # 155 Operator C:
Zero Equals: Specification: Area: QA room
Unit of Measure: mm Total Tolerance: Date:
Instructions:
1) Verify the gauge can read at least one additional degree of resolution than the part characteristic. Typically, a ten to one ratio of
tolerance to gauge resolution is required.
2) Verify the gauge is calibrated.
3) Perform study per section 5.1 of Instructions
4) Record results below. Type in shaded areas only. Note: "Total Tolerance" and "Operator Names" MUST be filled in for the form to
work properly.
5) Analyze the results to determine variability due to both Repeatability and Reproducibility.
Operator A: Operator B: Operator C:
Sample Trials Sample Trials Sample Trials
Number 1 2 3 R Number 1 2 3 R Number 1 2 3 R
1 1 1
2 2 2
3 3 3
4 4 4
5 5 5
6 6 6
7 7 7
8 8 8
9 9 9
10 10 10
Totals Totals Totals
Means Means Means
Xbar1 Xbar2 Xbar3 RbarA Xbar1 Xbar2 Xbar3 RbarB Xbar1 Xbar2 Xbar3 RbarC
XbarA = Xbar1 + Xbar2 + Xbar3 XbarB = Xbar1 + Xbar2 + Xbar3 XbarC = Xbar1 + Xbar2 + Xbar3
Trials Trials Trials
XbarA = XbarB = XbarC =
Test for Statistical Control of Ranges
Rbar1 = RbarA + RbarB + RbarC = ### = #VALUE!
Operators 2
UCLR = D4 * Rbar1 = 2.58 * #VALUE! = #VALUE! = Range Upper Control Limit
Note: D4 is based on the number of trials. (See table below)
Repeatability (Gage Variation)
GV = Rbar1 * K1 = #VALUE! * 3.05 = #VALUE!
Note: K1 is based on the number of trials. (See table below)
Reproduceability (Between Operator Variation)
Rbar2 = XbarLargest of ABC - XbarSmallest of ABC = 0.0000 - 0.0000 = 0.0000
Gauge R&R Long Form Study
OV = (Rbar2 * K2)2 - (GV2/(n*r)) = ( 0.000* 2.7)2-( 2
### / 0 * 0) = #VALUE!
Note: K2 is based on the number of Operators. (See table below), n is the number of parts, r is the number of trials
Combined Reproduceability and Repeatability (Measurement Variation)
R&R = OV2 + GV2 = ###2 + #VALUE!
2
= ###
Part Variation
PV = RP * K3 = #DIV/0! * 1.62 = #DIV/0!
Note: K3 is based on the number of parts. (See table below)
Total Variation
TV = R&R2 + PV2 = ###2 + #DIV/0!2 = ###
Table of Factors used in calculations from AIAG Manual for Measurement Systems Analysis, Chapter II, Section 4, Figure 19.
r D4 K1 Op. K2 n K3
2 3.27 4.56 2 3.65 9 1.67
3 2.58 3.05 3 2.7 10 1.62
% R&R (as a % of TV)
%R&R = R&R / TV = ### / #VALUE! = ###
THE SUM OF THE PERCENT CONSUMED BY EACH FACTOR WILL NOT EQUAL 100%
This result for percent of total GR&R Error shall be evaluated in order to determine if the measurement system as
intended is acceptable for its application.
Guidelines for Judging the acceptance of the GR&R Error are:
-- Under 10% error - the measurement system is acceptable.
-- 10% - 30% error - may be acceptable based on importance of application, cost of gauge, cost to repair, etc.
-- Over 30% error - Measurement system must be improved. Follow the corrective action process to identify the root
cause of the problem and have them corrected.
Note: If Gage R&R as a % of TV fails to meet the above criteria, but the Gage R&R as a % of Tolerance Range
does meet the critera, then the gage is acceptable.
% R&R (as a % of Total Tolerance Range)
%R&R = R&R / Total Tolerance = ### / = ###