IPC-TM-650 EN 2022 试验方法--.pdf - 第37页
Measurement Precision Study - Data Sheet ( continued) Condition C Reading Sample A verage 1 2 3 4 5 6 7 8 9 10 Condition D Reading Sample A verage 1 2 3 4 5 6 7 8 9 10 Condition E Reading Sample A verage 1 2 3 4 5 6 7 8 …
Beside each potential MSA candidate we could note information needed to help assess criticality. This information could
include the following:
• Issues identified as critical to customer (CTC) or quality (CTQ)
• Test cost
• Time required for test
• Unit cost of product being tested
• Planned production quantities
• Past test yields
• Cost to analyze test rejects
• Rework costs
• Scrap costs
• Warranty returns caused by issues with this test
Having this information, it is possible to then devise a prioritizing scheme that would identify critical test issues. Some have
used the planned production quantities and past test yields to calculate potential costs of rework and scrap. A different pri-
oritizing scheme might use knowledge of issues critical to the customer or past warranty claims.
The third step in preparing the MSA is to take the rank ordered list and prepare an action plan. Given the time and budget
allowed, decide on the tests to be included in the first round of MSAs. After the first group has been analyzed, one can then
consider the second group, etc.
The test method gives more detailed instructions. Some rules of thumb and general pieces of advice are discussed below.
Many MSAs are discussed, but never completed. There are several reasons. For some the procedure is confusing. Hopefully,
we have helped streamline it here. For others, the process is too time consuming and expensive. While it is true that more
samples and more testers will yield more precise estimates; nevertheless, an incomplete study is useless.
Therefore it is recommended that MSAs start small and efficient. Pick the highest priority test and design an MSA which
can be performed quickly with the time and resources available. If this initial study shows the need for improvement, then
call in an expert practitioner and design a more extensive study and begin corrective action.
In many cases, a good basic study may involve just five parts, two to three testers, and three or four measurements on each
part by each tester. Each tester could measure each part once a day for several days.
Begin with the highest priority test first, then continue with the rest. Many have found that a small effort continued over the
long term can accomplish more than a massive initiative. Develop a plan to reassess the MSA on a regular basis. It is
important to track the effect of the MSA efforts, both in term of the performance metric involved and in cycle time and cost
savings. Choose a representative group of samples for the MSA. Choose samples that cover the part type, range and set-up
under consideration. Be sure to use testers who represent those who will actually do the test during production.
Be sure the parts are presented to the testers in a manner which prevents them from knowing the prior measurements. This
prior knowledge may influence their current measurement and compromise the study.
Try to design a study so that all tests result in actual measurements. Try not to stress products beyond normal test limits,
so secondary effects do not occur. Do not set a part aside if it happens to result in a noncompliance. Continue taking mea-
surements. The estimates are more representative if the MSA is balanced, with equal numbers of test results all around.
Do not ignore any measurement taken, unless there was a special cause of variation that is clearly identified and preventa-
tive action has been taken.
2
Users
Guide
January
2003
Setting-Up
the
MSA
Measurement Precision Study - Data Sheet (continued)
Condition C Reading
Sample
Average1 2 3 4 5 6 7 8 9 10
Condition D Reading
Sample
Average1 2 3 4 5 6 7 8 9 10
Condition E Reading
Sample
Average1 2 3 4 5 6 7 8 9 10
Condition F Reading
Sample
Average1 2 3 4 5 6 7 8 9 10
IPC-TM-650
Page 7 of 10
Number
1.9
Subject
Measurement
Precision
Estimation
for
Variables
Data
Date
01/03
Revision
A
1
2
3
4
5
Average
Xc
=
Range
%
二
1
2
3
4
5
Average
X0
=
Range
R
。
=
1
2
3
4
5
Average
Range
1
2
3
4
5
Average
Range
Measurement Precision Study - Data Sheet (continued)
Condition G Reading
Sample
Average1 2 3 4 5 6 7 8 9 10
Condition H Reading
Sample
Average1 2 3 4 5 6 7 8 9 10
Condition I Reading
Sample
Average1 2 3 4 5 6 7 8 9 10
Condition J Reading
Sample
Average1 2 3 4 5 6 7 8 9 10
IPC-TM-650
Page 8 of 10
Number
1.9
Subject
Measurement
Precision
Estimation
for
Variables
Data
Date
01/03
Revision
A
1
2
3
4
5
Average
Xg
=
Range
rg
二
1
2
3
4
5
Average
X
修
=
Range
%
二
1
2
3
4
5
Average
X/
二
Range
R/
=
1
2
3
4
5
Average
Range