Method for calculating adaptive inference test figure of merit

• US 5,043,987 A
• Filed: 11/07/1989
• Issued: 08/27/1991
• Est. Priority Date: 11/07/1989
• Status: Expired due to Fees

• Current Assignee: Array Analysis Incorporated
• Original Assignee: Array Analysis Incorporated
• Inventors: Prestas, Gregory C., Stark, William D.
• Primary Examiner(s): Dixon, Joseph L.
• Application Number:
  US07/433,335
• Time in Patent Office: 658 Days
• Field of Search: 364/580, 364/579, 364/554, 364/513, 364/480, 364/481, 364/483, 364/188, 364/138, 371/21.4, 371/21.6, 371/22.1, 371/22.4, 371/23, 371/24, 371/25.1, 371/26, 371/29.1, 371/15.1, 324/77 R, 324/77 A, 324/114, 324/121 R, 324/158 R
• US Class Current: 714/737
• CPC Class Codes: G01R 31/00 Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere measuring superconductive properties G01R33/1238; testing or measuring semiconductors or solid state devices during manufacture H01L22/00; testing line transmission systems H04B3/46
• CPC Class Codes: Y10S 706/916 Electronic or computer, internal or network, circuit

Abstract

A method for using an adaptive inference system to detect and locate faults in an electrical or electronic device or assembly. The steps include performing a position-dependent, time-ordered test upon an electrical or electronic device or assembly to provide comprehensive error analysis. The error analysis includes a library of error information. Previously unobserved faults can be detected and located despite their possible remoteness within the device or assembly. When detected faults are compared with error library, a figure of merit is calculated for all possibilities of detected faults.

Claims

1. An adaptive inference method of calculating a figure of merit for detected faults in an electrical or electronic device or assembly, some of said faults not having been observed in prior tests, or obtained from stored data and information in said adaptive inference system, or bearing a direct relationship to a problem being reviewed, said method comprising the steps of:

a) performing a position-dependent, time-ordered test upon an electrical or electronic device or assembly to provide a comprehensive error analysis;

b) developing an error plane or detected faults consisting of channels on one axis of said error plane and time samples on another axis of said error plane, each channel of said error plane being represented in three ways;

(i) as bits, defining binary words representing a number of errors,

(ii) as groups, defining a binary word representing a number of times an error changes from a "no error condition" to an "error condition," and

(iii) as a range, defining bits that represent an error over a segment of said error plane;

(c) comparing said detected faults with stored faults data and information; and

(d) calculating a figure of merit for each of said bits, groups and range errors, and weighting and summing each of said bits, groups and range errors to provide a total figure of merit for said detected fault.

2. The adaptive inference method of calculating a figure of merit for detected faults in an electrical or electronic device or assembly in accordance with claim 1, wherein said bit errors and group errors are each weighted approximately 1/4, and said range errors are weighted approximately 1/2.

3. The adaptive inference method of calculating a figure of merit for detected faults in an electrical or electronic device or assembly in accordance with claim 1, wherein said figure of merit for each bit, group and range error is calculated according to the formula:

space="preserve" listing-type="equation">match bits/base bits×

match bits/new bits

wherein base bits are defined as the number of error bits in a stored error plane, new bits are defined as the number of bits of an acquired error plane, and match bits are defined as the number of bits in common between base bits and new bits.

4. The adaptive inference method of calculating a figure of merit for detected faults in an electrical or electronic device or assembly in accordance with claim 2, wherein said figure of merit for each bit, group and range error is calculated according to the formula:

space="preserve" listing-type="equation">match bits/base bits×

match bits/new bits

wherein base bits are defined as the number of error bits in a stored error plane, new bits are defined as the number of bits of an acquired error plane and match bits are defined as the number of bits in common between base bits and new bits.

5. The adaptive inference method of calculating a figure of merit for detected faults in an electrical or electronic device or assembly in accordance with claim 1, wherein fault data and information is transition encoded.

6. An adaptive inference method of calculating a figure of merit for detected faults in an electrical or electronic device or assembly, some of said faults not having been observed in prior tests, or obtained from stored data and information in said adaptive inference system, or bearing a direct relationship to a problem being reviewed, some method comprising the steps of:

a) performing a position-dependent, time-ordered test upon an electrical or electronic device or assembly to provide a comprehensive error analysis;

b) developing an error plane of detected faults consisting of channels on one axis of said error plane and time samples on another axis of said error plane, each channel of said error plane comprising serial bit streams representing a number of errors;

c) comparing said detected faults with stored fault data and information; and

d) calculating a figure of merit for each of said bit stream errors, and weighting and summing each of said bit stream errors to provide a total figure of merit for said detected fault.

7. The adaptive inference method of calculating a figure of merit for detected faults in an electrical or electronic device or assembly in accordance with claim 6, wherein said figure of merit for each bit stream is calculated according to the formula:

space="preserve" listing-type="equation">match bits/base bits×

match bits/new bits

wherein base bits are defined as the number of error bits in a stored error plane, new bits are defined as the number of bits of an acquired error plane, and match bits are defined as the number of bits in common between base bits and new bits.

8. The adaptive inference method of calculating a figure of merit for detected faults in an electrical or electronic device or assembly in accordance with claim 6, wherein said bit streams comprise J-Tag data.

9. The adaptive inference method of calculating a figure of merit for detected faults in an electrical or electronic device or assembly in accordance with claim 6, wherein said bit streams comprise boundary scan information.

10. The adaptive inference method of calculating a figure of merit for detected faults in an electrical or electronic device or assembly in accordance with claim 6, wherein fault data and information is transition encoded.