A scientific study of the problems
of digital engineering for space flight systems,
with a view to their practical solution.
Presented at the 1995 RADECS
Richard Katz
NASA Goddard Spaceflight Center
Greenbelt, MD, USA
Gary Swift, David Shaw
Jet Propulsion Laboratory
California Institute of Technology
Pasadena, CA, USA
Abstract
Gamma irradiation and annealing of a large number of Actel FPGAs with in-situ current measurements were performed. Lot-to-lot, part-topart, and burn in variations were measured. Findings include a catastrophic failure mechanism and minimal dose rate effects.
Table of Contents
I. Introduction
II. Test Methodology
A. Test Devices
B. Test ProgramIII. Test Results
A. Lot-To-Lot Variations
B. Feature Size
C. Radiation Bias
D. Effect of Charge Pump on Total Dose ResponseIV. Discussion
V. Conclusions
VI. References
List of Figures
Figure 1. Increase in Power Supply Current of the A1020B During Irradiation (Static Bias).
Figure 2. Increase in Power Supply Current of the A1280A During Irradiation (Static Bias).
Figure 3. Degradation of Propagation Delay Time of theA1020 Showing Increase After High-Temperature Annealing.
Figure 4. Effect of Total Dose on an A1280A Device Irradiated without Bias.
Figure 5. Comparison of A1020B Devices from Three Different Date Codes.
Figure 6. Comparison of the Radiation Response of A1020-Series Devices with Different Feature Size.
Figure 7. Effect of Static Bias Conditions on Operating Current for the Actel A1020B.
Figure 8. Dynamic vs. static bias comparison for the Actel A1020B.
Figure 9. Minimum Operating Voltage of the 1020B Irradiated Under Static and Dynamic Operating Conditions.
Figure 10. The Relationship between VCC and Operating Current of the A1020B at Different Radiation Levels.
Figure 11. Start-up Current Transients in the Actel 1280A.
Figure 12. Effect of a Biased Irradiation on a Device Previously Irradiated without Bias.
List of Tables
Table 1. Comparison of the Radiation Response of A1280A Devices with and without Burn In.
Conclusions
The A1020B is not nearly as tolerant to total dose as previous versions of this device with larger feature size, although substantial variations in radiation hardness occur between different production lots for all feature sizes. Total dose degradation affects the charge-pump circuit in these devices, causing large increases in the power supply current and raising the minimum operating voltage.
The improved charge pump used in the A1280 family increases its radiation susceptibility compared to the A1020-series devices. The A1280A can be expected to draw significant current (> 100 mA, static) after only a few krad(Si), well before functional failure at 10-15 krad(Si). Both device types show little part-to-part variability within a lot, but significant variations between lots. The effects of burn-in on the dose susceptibility, if any, are smaller than part-to-part variation. Irradiating an A1280A without bias significantly lowers the dose response, although it appears as if it may enhance the effects of subsequent biased irradiation. The charge pump damage of both FPGA families anneals readily. Thus, it is possible that low dose rate testing may show a significant decrease in susceptibility for the same total dose.
The large currents that are required during start up for these devices is a significant problem for many applications. It is vitally important that radiation testing include special measurements of the start-up characteristics, which are reported for the first time in this paper. The large currents not only increase power dissipation in irradiated devices, but delay the operation of the circuit for extended time intervals after power is applied. If the power supply system cannot supply sufficient current, the charge-pump circuit will never be able to turn on, and the circuit will fail in the application at lower radiation levels.
Home
Last Revised January 09, 2002
Digital Engineering Institute
Web Designer: Richard Katz