# US 2004/0064771 A1 · Method and system for coding test pattern for scan design

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2004/0064771 A1
- Document ID: US20040064771A1
- Status: Published application
- Kind: A1
- Application number: 10/630,957
- Filed: 2003-07-30
- Published: 2004-04-01
- Earliest claimed: 2002-07-30
- Assignees: International Business Machines Corporation
- Inventors: Joerg Georg Appinger; Michael Juergen Kessler; Manfred Schmidt
- Web page: https://patentgrep.com/patent/US20040064771A1

## CPC

- G01R 31/318335: {Test pattern compression or decompression (compression or decompression of scan patterns G01R31/318547; compression or decompression hardware G01R31/31921)}
- G01R 31/318385: {Random or pseudo-random test pattern}
- G01R 31/318544: {Scanning methods, algorithms and patterns (G01R31/3183 takes precedence)}
- G01R 31/318566: {Comparators; Diagnosing the device under test}
- G01R: MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES (indicating correct tuning of resonant circuits H03J3/12)

## Abstract

A method and system for efficiently coding test pattern for ICs in scan design and build-in linear feedback shift register (LFSR) for pseudo-random pattern generation. In an initialization procedure, a novel LFSR logic model is generated and integrated into the system for test data generation and test vector compression. In a test data generation procedure, test vectors are specified and compressed using the LFSR logic model. Every single one of the test vectors is compressed independently from the others. The result, however, may be presented all at once and subsequently provided to the user or another system for further processing or implementing in an integrated circuit to be tested. According to the present invention a test vector containing 0/1-values for, e.g., up to 500.000 shift registers and having, e.g., about 50 so called care-bits can be compressed to a compact pattern code of the number of care-bits, i.e., 50 bits for the example of 50 care-bits.

## Claims

### Claim 1

A method for coding test pattern for integrated circuits (ICs) in scan design and with build-in test hardware (BIT-HW), whereby the ICs have IC design data, the BIT-HW consists of a linear feedback shift register (LFSR) for pseudo-random pattern generation and a computer program for pattern merging and distribution over scan chains, the method comprising the steps of:
specifying a logic model representation of the physical BIT-HW;
specifying test vectors one by one;
compressing said test vectors one by one; and
providing the compressed test vectors.

### Claim 2 (depends on 1)

The method according to claim 1, whereby said step of specifying the logic model representation includes a step of specifying a function operator and an initial state-function from respective IC design data, whereby said function operator is an executable logic model representation of the compressed test vectors.

### Claim 3 (depends on 2)

The method according to claim 2, whereby the step of specifying the logic model representation includes a step of specifying a set of chain access operators from respective IC design data, and whereby each chain access operator corresponds to a certain scan chain and is an executable logic model representation of said means for pattern merging and distribution.

### Claim 4 (depends on 2)

The method according to claim 2, whereby the step of specifying the logic model representation includes a step of generating an LFSR generator code from said initial state-function by an iterative execution of said function operator.

### Claim 5 (depends on 2)

The method according to claim 2, whereby the step of specifying the logic model representation includes a step of generating state-functions building an LFSR generator matrix line by line beginning with said initial state-function by an iterative execution of said function operator.

### Claim 6 (depends on 4)

The method according to claim 4, whereby the step of specifying the logic model representation includes a step of generating state-functions building a set of chain LFSR generator matrices from said LFSR generator code each by iterative executions of the respective chain access operators.

### Claim 7 (depends on 3)

The method according to claim 3, whereby the step of specifying the logic model representation includes a step of generating state-functions building a set of chain LFSR generator matrices from said initial state-function each by iterative executions of said function operator in combination with the respective chain access operators.

### Claim 8 (depends on 4)

The method according to claim 4, whereby said LFSR generator code is a binary vector of a length that corresponds to a certain maximum number of execution cycles of said LFSR.

### Claim 9 (depends on 5)

The method according to claim 5, whereby the LFSR generator matrix is a binary matrix with a line width of the length of the LFSR and a column length of the length of the respective scan chain.

### Claim 10 (depends on 5)

The method according to claim 5, whereby said state-function is a binary vector having the length of the LFSR and represents an executable function to be executed by the following action sequence:
a. selecting bit positions of the LFSR in correspondence to the 1-values of the state-function;
b. combining the 0/1-values of the selected bit positions by XOR arithmetic; and
c. presenting the XOR-output 0/1-value to the user or another system for further processing.

### Claim 11 (depends on 1)

The method according to claim 1, whereby the step of specifying a test vector includes a step of specifying a set of care-bits from respective IC design data, specifying for each care-bit a 0/1-value and specifying a position referenced to a specific scan chain and execution cycle in which this 0/1-value is generated by the LFSR.

### Claim 12 (depends on 5)

The method according claim 5, whereby the step of compressing a test vector includes a step of selecting and providing a collection of state-functions out of the LFSR generator matrix, whereby the step of specifying a test vector includes a step of specifying a set of care-bits from respective IC design data, and whereby said specification of the respective care-bit position is the index in this selection.

### Claim 13 (depends on 4)

The method according to claim 4, whereby the step of compressing a test vector includes a step of computing and providing a collection of state-functions, whereby the step of specifying a test vector includes a step of specifying a set of care-bits from respective IC design data, and whereby said specification of a care-bit position is the index for selecting a sequence-vector out of the LFSR generator code and for selecting a chain access operator for this computing.

### Claim 14 (depends on 12)

The method according to claim 12, whereby said step of compressing a test vector includes a step of solving a linear equation system formed by said collection of state-functions, whereby the solution is a compressed LFSR-Code.

### Claim 15 (depends on 14)

The method according to claim 14, whereby the step of compressing a test vector and solving a linear equation system comprises a method for computing the LFSR-Code in a variable length code, whereby the method provides a special solution characterized by the longest left or right adjusted uniform sequence of either zeros or ones.

### Claim 16 (depends on 15)

The method according to claim 15, whereby the step of providing the compressed test vectors in a variable length code includes a step of sorting and grouping the total of compressed test vectors over the individual lengths and storing them in data records having suitable formats for each length interval.

### Claim 17 (depends on 1)

The method according to claim 1, whereby the step of providing the compressed test vectors includes a step of simulating a test execution using the compressed test vectors and providing the simulation results to the user or another system for further processing.

### Claim 18

A system for coding test pattern for integrated circuits (ICs) in scan design, said system comprising:
build-in test hardware (BIT-HW), whereby the ICs have IC design data;
a linear feedback shift register (LFSR) included in said BIT-HW for pseudo-random pattern generation; and
a computer system including instructions to execute a method for pattern merging and distribution over scan chains, said method comprising the steps of:
specifying a logic model representation of the physical BIT-HW;
specifying test vectors one by one;
compressing said test vectors one by one; and
providing the compressed test vectors.

### Claim 19 (depends on 18)

The system according to claim 18, whereby said method step of specifying the logic model representation includes a step of specifying a function operator and an initial state-function from respective IC design data, whereby said function operator is an executable logic model representation of the compressed test vectors.

### Claim 20 (depends on 19)

The system according to claim 19, whereby the method step of specifying the logic model representation includes a step of specifying a set of chain access operators from respective IC design data, and whereby each chain access operator corresponds to a certain scan chain and is an executable logic model representation of said means for pattern merging and distribution.

### Claim 21 (depends on 19)

The system according to claim 19, whereby the method step of specifying the logic model representation includes a step of generating an LFSR generator code from said initial state-function by an iterative execution of said function operator.

### Claim 22 (depends on 19)

The system according to claim 19, whereby the method step of specifying the logic model representation includes a step of generating state-functions building an LFSR generator matrix line by line beginning with said initial state-function by an iterative execution of said function operator.

### Claim 23 (depends on 20)

The system according to claim 20, whereby the method step of specifying the logic model representation includes a step of generating state-functions building a set of chain LFSR generator matrices from said LFSR generator code each by iterative executions of the respective chain access operators.

### Claim 24 (depends on 20)

The system according to claim 20, whereby the method step of specifying the logic model representation includes a step of generating state-functions building a set of chain LFSR generator matrices from said initial state-function each by iterative executions of said function operator in combination with the respective chain access operators.

### Claim 25

A program product for coding test pattern for integrated circuits (ICs) in scan design and with build-in test hardware (BIT-HW), whereby the ICs have IC design data, the BIT-HW consists of a linear feedback shift register (LFSR) for pseudo-random pattern generation, the program product comprising:
a computer readable medium having recorded thereon computer readable program code for performing a method of pattern merging and distribution over scan chains, said method comprising the steps of:
specifying a logic model representation of the physical BIT-HW;
specifying test vectors one by one;
compressing said test vectors one by one; and
providing the compressed test vectors.

### Claim 26 (depends on 25)

The program product according to claim 25, whereby said method step of specifying the logic model representation includes a step of specifying a function operator and an initial state-function from respective IC design data, whereby said function operator is an executable logic model representation of the compressed test vectors.

### Claim 27 (depends on 25)

The program product according to claim 25, whereby the step of specifying the logic model representation includes a step of specifying a set of chain access operators from respective IC design data, and whereby each chain access operator corresponds to a certain scan chain and is an executable logic model representation of said means for pattern merging and distribution.

### Claim 28 (depends on 25)

The program product according to claim 25, whereby the method step of specifying the logic model representation includes a step of generating an LFSR generator code from said initial state-function by an iterative execution of said function operator.

### Claim 29 (depends on 26)

The program product according to claim 26, whereby the method step of specifying the logic model representation includes a step of generating state-functions building an LFSR generator matrix line by line beginning with said initial state-function by an iterative execution of said function operator.

### Claim 30 (depends on 28)

The program product according to one of the claims 28, whereby the method step of specifying the logic model representation includes a step of generating state-functions building a set of chain LFSR generator matrices from said LFSR generator code each by iterative executions of the respective chain access operators.

### Claim 31 (depends on 27)

The program product according to one of the claims 27, whereby the method step of specifying the logic model representation includes a step of generating state-functions building a set of chain LFSR generator matrices from said initial state-function each by iterative executions of said function operator in combination with the respective chain access operators.
