Test system for detecting faults in multiple devices of the same type

A test system for testing a plurality of devices of a same type, comprising:

• a plurality of built-in self-test (BIST) devices, wherein each of the plurality of devices of the same type include one of the plurality of BIST devices;
• at least one processor;
• and
• at least one memory storing instructions, that in response to determining that executed by the at least one processor, cause the test system at least to:
• randomly generate a stimulus signal based on a stimulus parameter;
• apply, by the one of the plurality of BIST devices, the stimulus signal N times to the plurality of devices of the same type;
• measure, by the plurality of devices of the same type, a response of the plurality of devices of the same type to the stimulus signal to produce M×N response outputs, wherein M is a number of the plurality of devices of the same type;
• calculating a defect likelihood for a test set of the plurality of devices of the same type based upon a mean of a reference set of the plurality of devices of the same type response outputs, a mean of a test set response outputs, a standard deviation of reference set response outputs, and a standard deviation of a test set response outputs;
• determine that the defect likelihood for the test set is greater than a first threshold;
• apply an initial operation of a directed random search algorithm to obtain an updated stimulus parameter in response to determining that the defect likelihood is greater than the first threshold, wherein the updated stimulus parameter is used to generate updated stimulus signal applied to the plurality of devices of the same type to improve detection of a defect;
• apply, by the BIST devices, the updated stimulus signal N times to the plurality of devices of the same type;
• measure, by the plurality of devices of the same type, a response of the plurality of devices of the same type to the updated stimulus signal to produce M×N updated response outputs;
• calculating a defect likelihood for the test set based upon a mean of a reference set updated response outputs, a mean of a test set updated response outputs, a standard deviation of a reference set updated response outputs, and a standard deviation of a test set updated response outputs;
• and
• determining that the defect likelihood for the test set is greater than a second threshold, wherein the second threshold is greater than the first threshold.

A method of testing a plurality of devices of a same type wherein each of the plurality of devices of the same type include a built-in self-test device, comprising:

• generating, by a discriminator, a stimulus signal based on a stimulus parameter, wherein the discriminator is a machine learning model;
• applying, by the built-in self-test device, the stimulus signal to the plurality of devices of the same type;
• measuring a response of the plurality of devices of the same type to the stimulus signal to produce M response outputs, wherein M is a number of the plurality of devices of the same type;
• generating, by a generator, M−1 weights, wherein the generator is a machine learning model;
• calculating a first response output from a first identical device of the plurality of devices of the same type as a weighted average of the other M−1 response outputs using the M−1 weights associated with M−1 other devices of the same type;
• determining a difference between an estimated first response output and the first response output;
• updating the discriminator and the generator using the first response output in response to determining that the difference is less than a threshold;
• updating the discriminator and the generator using the estimated first response output in response to determining that the difference is greater than the threshold;
• and
• controlling, by a controller the discriminator and the generator to act as a generative adversarial network wherein the discriminator searches for stimulus parameter that indicate defects and wherein the generator to generate M−1 weights that indicate defects.

A test system for testing a plurality of devices of a same type wherein each of the plurality of devices of the same type include a built-in self-test device, comprising:

• a plurality of built-in self-test (BIST) devices, wherein each of the plurality of devices of the same type include one of the plurality of BIST devices;
• and
• at least one processor;
• and
• at least one memory storing instructions, that in response to determining that executed by the at least one processor, cause the test system at least to:
• generate stimulus parameters by a discriminator implemented as a machine learning model in the at least one processor;
• apply the stimulus parameters to the plurality of BIST devices;
• receive M response outputs from the plurality of BIST devices, wherein M is a number of the plurality of devices of the same type;
• generate M−1 weights by a generator implemented as a machine learning model in the at least one processor;
• estimate a first response output from a first identical device of the plurality of devices of the same type as a weighted average of the other M−1 response outputs using the M−1 weights associated with M−1 other devices of the same type;
• determine a difference between an estimated first response output and the first response output;
• update the discriminator and the generator using the first response output in response to determining that the difference is less than a threshold;
• update the discriminator and the generator using the estimated first response output in response to determining that the difference is greater than the threshold;
• and
• control the discriminator and the generator to act as a generative adversarial network wherein the discriminator searches for stimulus parameters that indicate defects and wherein the generator to generate M−1 weights that indicate defects.