GEIA EIA-591-1992 Assessment of Quality Levels in PPM Using Variables Test Data (Formerly TechAmerica EIA-591)《用变量试验数据评定PPM的质量等级》.pdf

《GEIA EIA-591-1992 Assessment of Quality Levels in PPM Using Variables Test Data (Formerly TechAmerica EIA-591)《用变量试验数据评定PPM的质量等级》.pdf》由会员分享，可在线阅读，更多相关《GEIA EIA-591-1992 Assessment of Quality Levels in PPM Using Variables Test Data (Formerly TechAmerica EIA-591)《用变量试验数据评定PPM的质量等级》.pdf（44页珍藏版）》请在麦多课文档分享上搜索。

1、EIA STANDARD ANSI/EIA-591- 1992 Approved: November 12, 1992 Reaffied: February 11,2000 Reaffirmed: September 5, 2002 Assessment of Quality Levels in PPM Using Variables Test Data EIA-591 DECEMBER 1992 ELECTRONIC INDUSTRIES ALLIANCE GOVERNMENT ELECTRONICS AND INFORMATION TECHNOLOGY ASSOCIATION A SECT

2、OR OF Elsctrlinic Indu6tries Alliance Copyright Government Electronics McGraw-Hill, New York, 1983. David, H.A., Order Statistics; John Wiley (ii) the specification includes only a lower limit (Section (iii) the specification includes only an upper limit (Section 8.); 9.) As indicated in Section l.,

3、 these procedures are applied to variables data. In the following example, imagine that the data values represent measurements of electrical resistance in ohms on twenty resistors that are intended to have a specification nominal resistance of 1,000 ohms.* The sample data, in ohms, are as follows: 1

4、016 1047 1055 1015 1028 960 1009 1011 1018 978 1021 992 996 978 977 1047 1021 1019 1040 1004 Now, using any computer program that calculates correlation coefficient, make two vertical columns. The first column is the sample data above, put in ascending order and labeled Y. The values for column X ar

5、e found in Table 2. For this example, 20 data values are used, so column X is a copy of the values under “Sample Size 20“. * Note that the data values could just as well represent measurements of tensile strength in pascals, or length in millimeters, or capacitance in farads, or measurements of any

6、other variable in appropriate physical units. Copyright Government Electronics it falls between 4.1967 and 4.9961. Consequently, it is necessary to interpolate between these two values which represent fractions nonconforming (P) of .O01 and .0001, as read across the top of Table 3. The K values of T

7、able 3 and 4 are different from the Z values found in the Area Table, so to reduce the chance of error, Table 5 gives the Z values corresponding to the fractions nonconforming (P) used in Tables 3 and 4. For this example, a K value of 4.1967 corresponds to a P value of .001. From Table 5, this corre

8、sponds to a Z value of 3.090. (Note that for Z = 3.090, the Area Table, Appendix B, gives an area of 0.4989992. Since Appendix B is for one-half of a normal distribution, subtract 0.4989992 from 0.5000000 to get 0.0010008, which is rounded off to 0.001 in Table 5). Similarly, Table 5 shows that P =

9、0.0001 corresponds to a Z value of 3.719. Now, use linear interpolation to find the value of Z1 that corresponds to the calculated value of K1. P z K o. 001 o. 0001 3.090 4.1967 3.719 4.9961 21 4.3370 (Kl) Z1 = 3.090 + (4.3370 - 4.1967) * (3.719-3.090) (4.9961 - 4.1967) 21 = 3.200 Now, look up this

10、value of 21 in the Area Table (Appendix B) to find the equivalent area under the normal curve, which is equal to -0.499313. The resulting fraction nonconforming, expressed in PPM is: PPM1 = (0.500000-0.499313) * 106 = 687 PPM Copyright Government Electronics however, this shorter statement may often

11、 be the desired form of the conclusion. Note the difference between this value of 6996 PPM and the tltypicaltl value of 302 PPM naively calculated earlier. Remember that this conclusion deDends critically on the underlvinq poriulation distribution beina normal. 8. ESTIMATION OF PPM NONCONFORMING TO

12、A LOWER SPECIFICATION LIMIT The procedure for estimating quality level relative to a one-sided Copyright Government Electronics pages 276 - 298, Parts Per Million Values For Estimatina Qualitv Levels, by courtesy of Marcel Dekker Inc. (1988). Copyright Government Electronics pages 106 - 128, Parts P

13、er Million Values For Estimatina O ualitv Levels, by courtesy of Marcel Dekker Inc. (1988) . Copyright Government Electronics do not group the data into cells. Tied values also should be treated individually. For Copyright Government Electronics National Bureau of Standards Applied Mathematics Serie

14、s 23; June 5, 1953. * Abramowitz, M. and Stegun, I. A.; Handbook of Mathematical Functions with Formulas. GraDhs. and Mathematical Tables. National Bureau of Standards, Applied Mathematics Series 55, June 1964, p. 933, equation 26.2.23. Copyright Government Electronics & Information Technology Association Reproduced by IHS under license with GEIA Not for ResaleNo reproduction or networking permitted without license from IHS-,-,-