How true to the specification is this ASIC likely to perform throughout its lifetime? Whether NASA flies an ASIC depends on the answer to that question. This section presents ways to assure that a flight ASIC meets all requirements, especially those that can be explicitly contracted. The approach described here represents state- of-the-practice methodology and is the best approach known today to assure success in an ASIC's final application. It includes a combination of complete requirement definition and specification, conscientious supervision, informed design, and proper part acceptance.
ASIC part acceptance requires a verification approach different from off-the-shelf devices. Off-the-shelf devices have reliability data bases compiled from previous applications. This data base provides reliability assurance that reduces or eliminates the need for verification tests. In contrast, an ASIC design has little or no previous applications and therefore lacks significant reliability history. Therefore, unlike off-the-shelf parts, ASICs require extensive verification under all anticipated conditions. Since the user is the designer, the user must be deeply involved in this verification process by specifying tests and screens, supplying tests vectors, providing test diagnosis, and verifying engineering (prototype) parts.
This section focuses on those activities that constitute traditional off- the-shelf part acceptance, with additional material for ASIC devices. However, before we address this material we will identify part acceptance work that occurs as subordinate activity in other sections of this guide.
The part acceptance process provides no guarantees. Part acceptance is a set of commonly accepted activities that verify each part produced can pass certain contracted tests. These end-of-the-line tests may reveal anomalous defects but they do not reveal every possible failure mechanism to which the part may be vulnerable throughout its lifetime. Fabrication process quality directly relates to the presence and severity of failure mechanisms. Therefore, we cannot overstate the importance of choosing a vendor who has a qualified process line.
Government-qualified services greatly reduce the cost of part acceptance by generically qualifying vendors to produce many types of parts. This alleviates most of the qualification expense that would be required per part without government qualification.
The appendix "Government Qualification Programs" has detailed discussions of these programs and how they relate to part acceptance.
"The user must be deeply involved in the verification process, specifying tests and screens, supplying test vectors, providing test diagnosis, and verifying engineering parts."
The terms "quality" and "reliability" have vague meanings in government standards and the integrated circuit industry. For clarity, we define these concepts as follows. "Quality": the extent to which the user organization gets what it requires explicitly in its part contract and implicitly in the vendor's in-house specifications. "Reliability": the likelihood of a part to operate according to its specification throughout its specified lifetime.
"Part acceptance ensures parts with high quality and reliability."
Reliability calls for prediction, making its requirements more difficult to assess than quality, which only requires description. But these predictions determine whether a part will be accepted for flight. The vendor and sometimes the user apply many tests that yield information that helps determine how long a part is likely to last. These tests, along with test structure measurements, provide the necessary data to make reliability predictions.
We discuss how tests and test structures provide predictive data in the appendix "Reliability".
Now you may jump to: