Method for testing electronic assemblies in the presence of noise

• US 5,327,437 A
• Filed: 11/25/1991
• Issued: 07/05/1994
• Est. Priority Date: 11/25/1991
• Status: Expired due to Fees

• Current Assignee: Agilent Technologies, Inc.
• Original Assignee: Hewlett-Packard Company (HP Inc.)
• Inventors: Balzer, Raymond J.
• Primary Examiner(s): Baker, Stephen M.
• Application Number:
  US07/797,616
• Time in Patent Office: 953 Days
• Field of Search: 371/25.1, 364/580, 364/554, 364/575
• US Class Current: 714/736
• CPC Class Codes: G01R 31/01 Subjecting similar articles in turn to test, e.g. "go/no-go" tests in mass production; Testing objects at points as they pass through a testing station testing of cables continuously passing the testing apparatus G01R31/59; testing dielectric strength or breakdown voltage G01R31/12
• CPC Class Codes: G01R 31/2846 using hard- or software simulation or using knowledge-based systems, e.g. expert systems, artificial intelligence or interactive algorithms

Abstract

An iterative test method which is operable in the presence of measurement noise uses the result of a particular measurement on a device to decide whether to pass or fail the-device or to take another measurement. The pass/fail decision is made by computing certain statistical values which "compare" the measurement result to a predefined pass/fail limit. Another test measurement is made if the result of the present measurement is too "close" to the pass/fail limit to make an accurate decision. If the result is not too "close" to the limit then a pass/fail decision is made based on whether the measurement is on the pass or fail side of the limit. The statistical values used to compare the measurement value to the limit can be Z-statistics or Student-t statistics depending on the test environment and the noise characteristics. Overall feedback can be added to dynamically adjust the statistical values in order to bring overall error rates into alignment with predetermined error rate targets.

Claims

1. A method for pass/fail testing an electronic assembly with a desired confidence target level by measuring a parameter in the presence of noise with known noise characteristics and comparing said measurement with a predetermined limit, said method comprising the steps of:

A. calculating at least one statistical limit value from said confidence target level;

B. obtaining a test value which represents a measured value of said parameter;

C. calculating an average measurement value from all test values of said parameter including the test value obtained in step B;

D. calculating a statistical value from said predetermined limit, said average measurement value calculated in step C and said noise characteristics;

E. repeating steps B through D when said statistical value calculated in step D has a predetermined relationship to said statistical limit value calculated in step A;

F. resolving the pass/fail status of said assembly by comparing said average measurement value to said predetermined limit when said statistical value calculated in step D has a relationship other than said predetermined relationship to said statistical limit value calculated in step A.

2. A method for pass/fail testing an electronic assembly according to claim 1 wherein step B comprises the step of:

B1. obtaining a test value by performing a single measurement of said parameter.

3. A method for pass/fail testing an electronic assembly according to claim 1 wherein step B comprises the steps of:

B2. obtaining a test value by performing a plurality of measurements of said parameter and mathematically combining said plurality of measurements.

4. A method for pass/fail testing an electronic assembly according to claim 1 wherein step A comprises the step of:

A2. calculating a statistical Z-value limit from said confidence target level by determining the number of standard deviations necessary to achieve said confidence target level from said normalized distribution curve.

5. A method for pass/fail testing an electronic assembly according to claim 1 further comprising the steps of:

G. calculating output error rates for said assembly;

H. comparing calculated error rates against predetermined error rate targets; and

I. adjusting said statistical limit value to cause said calculated error rates to approach said predetermined error rate targets.

6. A method for pass/fail testing an electronic assembly according to claim 1 wherein step A comprises the steps of:

A1. counting the number of measurements taken; and

A2. calculating a statistical Student-t value limit from said confidence target level and the number of measurements counted in step A1.

7. A method for pass/fail testing an electronic assembly according to claim 1 wherein step D comprises the step of:

D1. calculating a statistical Student-t value from said from said predetermined limit, said average measurement value calculated in step C and an estimate of the standard deviation of said noise.

8. A method for pass/fail testing an electronic assembly according to claim 1 wherein step D comprises the step of:

D1. calculating a statistical Z-value from said from said predetermined limit, said average measurement value calculated in step C and an estimate of the standard deviation of said noise.

9. A method for pass/fail testing an electronic assembly according to claim 8 wherein step D1 comprises the step of:

D1A. calculating a statistical Z-value from said from said predetermined limit, said average measurement value calculated in step C and an estimate of the standard deviation of said noise according to the formula;

##EQU4## where ABS is an absolute value operator, S is the estimated standard deviation and SQR is a square root operator.

10. A method for pass/fail testing an electronic assembly according to claim 8 wherein step D1 comprises the step of:

D1A. calculating a statistical Student-t value from said from said predetermined limit, said average measurement value calculated in step C and an estimate of the standard deviation of said noise according to the formula;

##EQU5## where ABS is an absolute value operator, S is the estimated standard deviation and SQR is a square root operator.

11. A method for testing an electronic assembly by measuring a parameter in the presence of noise with known noise characteristics, said method comprising the steps of:

A. establishing an acceptable range of values for said parameter;

B. establishing a desired confidence target for passing said assembly and a desired confidence target for failing said assembly;

C. calculating a first statistical limit value from said confidence target for passing said assembly and calculating a second statistical limit value from said confidence target for failing said assembly;

D. estimating the standard deviation of said noise;

E. measuring a value of said parameter;

F. calculating an average measurement value from all measured values of said parameter including the value measured in step E;

G. calculating a first statistical value and a second statistical value from said acceptable range of values established in step A, said average measurement value calculated in step F and said estimated standard deviation from step D;

H. comparing said first and said second statistical values calculated in step G to said first and said second statistical limit values calculated in step C;

I. repeating steps E through H when said first statistical value or said second statistical value calculated in step G is less than said first statistical limit value calculated in step C and said both first statistical value and said second statistical value are greater than said second statistical limit value calculated in step C;

J. passing said assembly when said first statistical value and said second statistical value calculated in step G are greater than said first statistical limit value calculated in step C; and

K. failing said assembly when said first statistical value or said second statistical value calculated in step G are less than said second statistical limit value calculated in step C.

12. A method for pass/fail testing an electronic assembly according to claim 11 wherein step C comprises the steps of:

C1. obtaining a normalized distribution curve; and

C2. calculating a first statistical Z-value limit from said confidence target for passing said assembly by determining the number of standard deviations necessary to achieve said confidence target level from said normalized distribution curve; and

C3. calculating a second statistical Z-value limit from said confidence target for failing said assembly by determining the number of standard deviations necessary to achieve said confidence target level from said normalized distribution curve.

13. A method for pass/fail testing an electronic assembly according to claim 11 wherein step G comprises the step of:

G1. calculating a first statistical Z-value from an upper limit of said acceptable range of values, said average measurement value calculated in step F and an estimate of the standard deviation of said noise according to the formula;

##EQU6## where S is the estimated standard deviation and SQR is a square root operator.

14. A method for pass/fail testing an electronic assembly according to claim 11 wherein step G comprises the step of:

G2. calculating a second statistical Z-value from a lower limit of said acceptable range of values, said average measurement value calculated in step F and an estimate of the standard deviation of said noise according to the formula;

##EQU7## where S is the estimated standard deviation and SQR is a square root operator.

15. A method for pass/fail testing an electronic assembly according to claim 11 further comprising the steps of:

L. calculating output error rates for said assembly;

M. comparing calculated error rates against predetermined error rate targets; and

N. adjusting said first and said second statistical limit values to cause said calculated error rates to approach said predetermined error rate targets.

16. A method for pass/fail testing an electronic assembly according to claim 11 wherein step D comprises the steps of:

D1. repeatedly measuring a single electronic assembly to obtain a plurality of measurements; and

D2. calculating an estimate of the standard deviation of from said plurality of measurements obtained in step D1.

17. A method for pass/fail testing an electronic assembly according to claim 11 wherein step I comprises the steps of:

I1. establishing a predetermined maximum number of measurements;

I2. repeating steps E through H when said first statistical value or said second statistical value calculated in step G is less than said first statistical limit value calculated in step C and said both first statistical value and said second statistical value are greater than said second statistical limit value calculated in step C and the number of measurements is less than or equal to the maximum number of measurements established in step I1; and

I3. discontinuing testing when the number of measurements is greater than the maximum number of measurements established in step I1.

18. A method for pass/fail testing an electronic assembly according to claim 17 wherein step I3 comprises the step of:

I3A. designating said electronic assembly as passed when the number of measurements is greater than the maximum number of measurements established in step I1.

19. A method for pass/fail testing an electronic assembly according to claim 17 wherein step I3 comprises the step of:

I3A. designating said electronic assembly as failed when the number of measurements is greater than the maximum number of measurements established in step I1.

20. A method for pass/fail testing an electronic assembly according to claim 17 wherein step I3 comprises the steps of:

I3A. designating said electronic assembly as marginal when the number of measurements is greater than the maximum number of measurements established in step I1.

21. A method for pass/fail testing an electronic assembly with a desired confidence target level by measuring a parameter in the presence of noise with known noise characteristics and comparing said measurement with a predetermined limit, said method comprising the steps of:

A. calculating a statistical limit value from said confidence target level;

B. selecting a measurement technique from a first measurement technique and a second measurement technique;

C. measuring a value of said parameter by means of a measurement technique selected in step B;

D. calculating an average measurement value from all measured values of said parameter including the value measured in step B;

E. calculating a statistical value from said predetermined limit, said average measurement value calculated in step C and said noise characteristics;

F. repeating steps B through E when said statistical value calculated in step E is less than said statistical limit value calculated in step A; and

G. resolving the pass/fail status of said assembly by comparing said average measurement value to said predetermined limit when said statistical value calculated in step E is greater than or equal to said statistical limit value calculated in step A.

22. A method for pass/fail testing an electronic assembly according to claim 21 wherein step B comprises the steps of:

B1. determining the cost of said first measurement technique and the cost of said second measurement technique;

B2. determining the measurement value variance of said first measurement technique and said second measurement technique;

B3. forming the product of the cost and the variance of said first measurement technique;

B4. forming the product of the cost and the variance of said second measurement technique; and

B5. comparing the product formed in step B3 to the product formed in step B4.

23. A method for pass/fail testing an electronic assembly according to claim 22 wherein step B further comprises the steps of:

B6. selecting said first measurement technique for a predetermined number of tests; and

B7. selecting said second measurement technique for all tests beyond said predetermined number of tests.

24. A method for pass/fail testing an electronic assembly according to claim 23 wherein step B further comprises the step of:

B8. adjusting said average measurement value calculated in step D when said second measurement technique is selected.

25. A method for pass/fail testing electronic assemblies with a desired confidence target level by making repeated measurements of at least one test parameter in the presence of noise and comparing said measurements with at least one predetermined pass/fail limit, said method comprising the steps of:

A. selecting a measurement method for measuring said test parameter which produces a set of measurement values in which the mean value of said set of measurement values approximates a true value of said test parameter;

B. obtaining a set of measurements by making repeated measurements of said test parameter on at least two electronic assemblies;

C. calculating an estimated standard deviation from said set of measurements obtained in step B;

D. selecting a statistical test method based on predetermined characteristics of said estimated standard deviation calculated in step C;

E. calculating at least one statistical limit value appropriate to the statistical test method selected in step D by using said confidence target level;

F. making at least one measurement of a value of said test parameter on a selected electronic assembly using the measurement method selected in step A;

G. calculating a weighted average measurement value from all measurements of said test parameter including the test value obtained in step F;

H. calculating a statistical value from said predetermined limit and said weighted average measurement value calculated in step G;

I. repeating steps F through H when said statistical value calculated in step H has a predetermined relationship to said statistical limit value calculated in step E; and

J. resolving the pass/fail status of said electronic assembly by comparing said average measurement value to said predetermined limit when said statistical value calculated in step H has a relationship other than said predetermined relationship to said statistical limit value calculated in step E.

26. A method for pass/fail testing electronic assemblies according to claim 25 wherein step A further comprises the step of:

A1. selecting a measurement method for measuring said test parameter which produces a set of measurement values in which the standard deviation of said set of measurement values is substantially constant.

27. A method for pass/fail testing electronic assemblies according to claim 25 wherein step A further comprises the step of:

A2. selecting a first measurement method for measuring said test parameter;

wherein step I comprises the step of;

I1. repeating steps F through H using said first measurement method selected in step A2 when said statistical value calculated in step H has a predetermined relationship to said statistical limit value calculated in step E for a predetermined number of repetitions;

wherein step A further comprises the step of;

A3. selecting a second measurement method for measuring said test parameter after said predetermined number of repetitions has been reached in step I1; and

wherein step I further comprises the step of;

I2. repeating steps F through H using said second measurement method selected in step A3 when said statistical value calculated in step H has a predetermined relationship to said statistical limit value calculated in step E.

C3. calculating a noise distribution from each set of measurements obtained in step B1 when said standard deviations are not substantially constant from electronic assembly to electronic assembly.

28. A method for pass/fail testing electronic assemblies according to claim 25 wherein step D comprises the steps of:

D1. selecting a Z-statistical test method when said estimated standard deviation calculated in step C is substantially constant from electronic assembly to electronic assembly; and

D2. selecting a Student-t statistical test method when said estimated standard deviation calculated in step C is not substantially constant from electronic assembly to electronic assembly.

29. A method for pass/fail testing electronic assemblies according to claim 25 wherein step E comprises the steps of:

E1. calculating at least one Z-statistical limit value from said confidence target level when a Z-statistical test method is selected in step D; and

E2. calculating a plurality of Student-t statistical limit values from said confidence target level when a Student-t statistical test method is selected in step D.

30. A method for pass/fail testing electronic assemblies according to claim 25 wherein step F comprises the step of:

F1. making a single measurement of said one value of said test parameter.

31. A method for pass/fail testing electronic assemblies according to claim 25 wherein step F comprises the step of:

F2. making two measurements of said one value of said test parameter and averaging the results of said two measurements to generate a measurement value.

32. A method for pass/fail testing electronic assemblies according to claim 25 wherein step G comprises the step of:

G3. adding the values of all measurements taken in step F to generate a measurement sum and dividing said measurement sum by a number of measurements taken in step F.

33. A method for pass/fail testing electronic assemblies according to claim 25 wherein step H comprises the steps of:

H1. calculating a statistical value from said predetermined limit and said weighted average measurement value calculated in step G and a pre-computed value for the standard deviation of the measurement noise when the statistical test method chosen in step D is the Z-statistical test; and

H2. calculating a statistical value from said predetermined limit and said weighted average measurement value calculated in step G and a value for the standard deviation of the measurement noise computed from measurement values obtained in step F when the statistical test method chosen in step D is the Student-t statistical test.

34. A method for pass/fail testing electronic assemblies according to claim 25 wherein step J comprises the step of:

J1. assigning an electronic assembly under test to a marginal category when the pass or fail status cannot be resolved.

35. A method for pass/fail testing electronic assemblies according to claim 25 wherein step B comprises the steps of:

B1. obtaining a set of measurements by making repeated measurements of said test parameter on a first electronic assembly; and

B2. replacing said first electronic assembly with a second electronic assembly different from said first electronic assembly; and

B3. repeating steps B1 and B2 a predetermined number of times.

36. A method for pass/fail testing electronic assemblies according to claim 35 wherein step C comprises the steps of:

C1. calculating an estimated standard deviation from each set of measurements obtained in step B1;

C2. comparing the estimated standard deviations calculated in step C1 to determine whether said standard deviations are substantially constant from electronic assembly to electronic assembly; and

C3. calculating a noise distribution from each set of measurements obtained in step B1 when said standard deviations are not substantially constant from electronic assembly to electronic assembly.

37. A method for pass/fail testing electronic assemblies according to claim 25 wherein step F comprises the step of:

F3. making a predetermined number of measurements of said one value of said test parameter and averaging the results of said two measurements to generate a measurement value.

38. A method for pass/fail testing electronic assemblies according to claim 37 wherein step G comprises the steps of:

G1. weighting each measurement value determined in step F3 by said predetermined number of measurements; and

G2. adding the values of all weighted measurements determined in step G1 to generate a measurement sum and dividing said measurement sum by a number of measurements taken in step F.