RTAX-S SEE High-Speed Effects

1. RTAX-S SEE High-Speed Effects Melanie D. Berg/MEI Ray Ladbury and Kenneth A. LaBel NASA/GSFC This work was sponsored by the NASA Electronic Parts and Packaging (NEPP) Program

2. Outline • Introduction • Radiation Tests Performed • Test Approach • Results • Future Tests

3. Introduction • Some preliminary single event effect (SEE) data existed on the RTAX-S devices • Lots during device development • Low-speed tests • Data showed fairly good single event upset (SEU) tolerance, but further data required • NEPP objectives for testing were to evaluate frequency, fanout, and combinatorial logic effects for both heavy ion and proton SEE • Testing performed at Texas A&M University (TAMU) Cyclotron on Nov 3-6 for heavy ions • Testing performed at Crocker Nuclear Laboratory (CNL) at the University of California at Davis (UCD) on Dec 9 for protons • All testing performed in collaboration with Actel

4. How We Tested • 5 Designs Under Test (DUT) Architectures were implemented • Designs were chosen to exemplify Frequency and Combinatorial-Logic Cross Section Effects under a variety of Ion Energies • The number of DFFs were not increased within each design (all strings consisted of 2000 DFFs). However, the combinatorial cross section varied. • Two examples of the DUT strings are below. • Test speeds were 15 MHz to 150 MHz 0F0L String - 0 levels of Combinatorial Logic with No Enable Logic 4F8L String - 8 levels of Combinatorial Logic with fan out of 4 within the Enable Control

5. RTAX-S TAMU Test Results • SEE Test Results Augmented Previous Actel Data • SEU sensitivity scales linearly with speed • Affects cross-section and LETth • Even “hardened” (triple-modular redundant – TMR) cells affected • SEU sensitivity increases with increased use of combinatorial logic • Single event transients (SETs) propagate • SEU sensitivity with fanout – TBD • Non-traditional SEU effects noted • Current increases noted with number of particles used for irradiation in every test run • Soft reset and power cycle cleared • Consistent with Actel data sets • See Actel’s presentation in this Briefing for detailed discussion of data, root cause (back end software), and fix. Cross-section scales with speed and combinatorial logic LETth scales with speed (LETth<<10) Sample Current Increase During Test Run

6. Best Case Results –No Combinatorial Logic

7. Worst Case Results at 150 MHz –Effect of Combinatorial Logic

8. Best Case Results with theEffect of Combinatorial Logic – 15 MHz

9. Calculated SEU Rates –CREME96 SEU rates at worst-case GEO In errors/bit-day RTSX-SU devices were also tested: similar results for increase in error rates with speed

10. Proton Test Results • Limited tests performed (lack of samples and time) • 0F0L at 15 MHz (best case) • 4F8L at 150 MHz (worst case) • Proton doses (63 MeV protons) of 100 to 200 krads(Si) per device • No SEUs observed at 15 MHz for any string • Few SEUs observed at worst case 150 MHz • SEU Cross-section of 1.33E-12 cm2 per device for 4F4L • SEU Cross-section of 7E-12 cm2 per device for 4F8L • Roughly an order of magnitude confidence level on results (low statistics) • No SEUs observed on other DUT strings at 150 MHz

11. Next Tests • Heavy Ion test scheduled for Feb (TBD as to exact dates) • Focus on • Verifying Actel “correction” for current anomaly • Testing internal RAM • I/O Transient (possible test, but not definite) • Further proton tests TBD • Acknowledgment • Thanks to Actel for providing devices, daughterboards, and support • We look forward to continued collaboration