A scientific study of the problems
of digital engineering for space flight systems,
with a view to their practical solution.
Presented at the 1995 International Test Conference, Workshop on IDDQ
Washington, D.C.
Richard B. Katz
NASA/Goddard Space Flight Center
Code 738
Greenbelt, MD 20771
Gary M. Swift
Jet Propulsion Laboratory
California Institute of Technology
4800 Oak Grove Drive, MS: 303-220
Pasadena, CA 91109
Abstract
A procedure for diagnosing faults using IDDQ techniques can identify leakage paths to power and ground as well as bridging faults. This diagnostic has been successfully applied to gate arrays with radiation-induced multiple faults. The feasibility of automating the procedure has been demonstrated and the underlying concepts appear compatible with certain design-for-test techniques.
Table of Contents
I. Introduction
II. Test Device
III. Fault Description
IV. Application of IDDQ Techniques
V. Implementation
VI. Conclusion
List of Figures
Figure 1. IDDQ Results for A1280A s/n:004. (Note: vector index 0 corresponds to all ones for walking zero and all zeros for walking one.) The final algorithm exploited the relationship between the walking 1’s and the walking 0’s data. By analyzing conditions that can cause IDDQ spikes, it is seen that an ’exclusive OR’ of the curves provides the distinction between shorts to rails and bridging faults. Corresponding (same vector index) spikes of the same polarity indicate that a change in that flip-flop's state relative to all the other flip-flops in the shift register cause the change in current. Corresponding spikes of opposite polarity indicates that the state of that individual flip-flop controls the anomalous current for that part of the DUT. This algorithm is easily automated and is efficient, needing only 2N+2 measurments. Lastly, pairs of vectors involved in the same bridging fault are identified by running exhaustive pairs tests on the minimal set of vectors known to be involved in bridging faults. The paper presents a detailed example making this vector selection and pair identification process clear.
Conclusion
Diagnosis of failed, functional VLSI devices with multiple faults is difficult. The procedures described here efficiently solve the problems of failure analysis, fault enumeration, and diagnosis. N vectors provide information for stimulus patterns to use in conjunction with diagnostic instruments such as emission microscopes. 2N+2 vectors are needed for fault enumeration which allowed us to make an accurate prediction of mission failure rates. Diagnosing the faults (type and matching vectors involved in bridging pairs) is done with a vastly reduced set of vectors tested pairwise, which initially was a combinatorially large problem. These techniques lend themselves to automation and have been implemented using only standard, inexpensive laboratory instrumentation. Lastly, design-for-testability techniques incorporated into VLSI devices and circuit boards can take advantage of these methods.
Home
Last Revised January 09, 2002
Digital Engineering Institute
Web Designer: Richard Katz