While the reliability of programmable logic devices is quite good, they do have a measurable failure rate. A database of device failures is being established. This will enable us to measure field reliability, categorize the failure modes, and spot trends as early as possible.
Failures will be divided into three classes:
- System software
- Design error
The first group, hardware failure, will obviously include devices that fail as a result of a manufacturing defect, as an example. Failures because of system application, such as an over voltage transient or ESD, should also be included. Programming failures for antifuse-based devices are not counted as failures, if the error is detected by the programmer's diagnostics. Programming yield information is also desired.
The second group of failures will include front-end software such as logic synthesizers, optimizers and translators, and back-end software such as place and route tools, simulation models, or timing analyzers.
The last group, design error, should include things such as logic design error, specification error, application error, assembly error, etc.
We are asking aerospace engineers to send in information about failures that they observe; all submissions will be kept anonymous, if desired. You may e-mail your information to firstname.lastname@example.org. Please include the following information, at minimum:
- Part Model Number
- Part Date and Lot Code
- Contact Name and Information
- Type of Failure
- Reason for Failure
- Supporting information, as available
Home - NASA
Office of Logic Design
Last Revised: February 03, 2010
Digital Engineering Institute
Web Grunt: Richard Katz