Radiation-Hardened re-programmable Field-Programmable Gate Array (RHrFPGA)

1
Radiation-Hardened re-programmable
Field-Programmable Gate Array (RHrFPGA)

• A.B. Sanders1, K.A. LaBel1, J.F. McCabe1,
• G.A. Gardner2, J. Lintz2, C. Ross2, K. Golke2,
• B. Burns2, M.A. Carts3, and H.S. Kim4
• NASA/GSFC, Code 561.4 Greenbelt, MD 20771
• Honeywell, Defense and Space Electronics Systems,
  Clearwater, FL 33764-7290
• Raytheon/ITSS, Lanham, MD 20706
• Jackson and Tull, Chartered Engineers, Seabrook,
  MD 20706

2
OUTLINE

• Introduction
• Radiation Test Suite
• Test Configuration
• Program Test Methods
• Test Procedure
• Test Results
• Summary
• Acknowledgements

3
INTRODUCTION

• Configurable Logic Blocks provide functional
  elements for constructing users logic
• I/O Cells provide the interface between the
  package pins and internal signal lines
• Programmable Interconnect Resources provide
  routing paths to connect the inputs and outputs
  onto the appropriate networks
• Customized configuration is established by
  programming internal static memory cells that
  determine the logic functions and internal
  connections implemented in the FPGA

INTERNAL FPGA
4
DEVICE CHARACTERISTICS

• Characteristics
• Silicon Epi thickness is 0.2 ?m
• Gate length is 0.35 ?m
• Material thickness over the Silicon Epi
  equivalent to 8.3 ?m(Si)
• Physical cross-sections of the memory and
  flip-flop cells are 0.45 ?m2 and 2.1 ?m2,
  respectively
• SOI technologies offer greater speed and power
  reduction compared to conventional bulk CMOS
• Motivation Why Develop the RHrFPGA?
• Reconfigurable FPGAs offer a significant
  advantage over one-time programmable Antifuse
  FPGAs
• Cache Logic design allows part of the FPGA to be
  reprogrammed without loss of register data, while
  the remainder continues to operate without
  disruption
• Rad hard need for space and military applications
  without complex external mitigation circuits
  (single chip solution)

5
BACKGROUND

• BACKGROUND Development of the RHrFPGA
• In July 1998, Honeywell and Atmel announced the
  signing of a license for Honeywell to develop a
  radiation hardened version of Atmels 30,000
  gate, 6400 register Field Programmable Gate Array
  (FPGA), the AT6010
• Honeywell developed a CMOS Silicon On Insulator
  (SOI) version of the AT6010 to meet the radiation
  hardness levels required for commercial and
  military space and missile systems
• The radiation hardened FPGA development was
  funded and managed by NASA Goddard Space Flight
  Center

6
RADIATION TEST SUITE
Interface Cable
PC
RTU Test Board

• Suite includes PC, Cables, Custom Remote
  Terminal Unit (RTU) Interface Dongle (in Cable),
  and RTU Test Board (Sockets for RHrFPGA DUT,
  foreground)

7
MULTI-APPLICATION
RHrFPGA Device Under Test Socket (under Beam)
Clock Input
Control RHrFPGA Socket (exercises DUT)
Program Command Input
Error Output
Power Input
Parallel Serial EEPROM Sockets for (opt.)
On-card Config.
Configuration Select Switches
8
ION BEAM CHARACTERISTICS
Ion Energy Angle Range (?m) Effective LET (MeV/(mg/cm2))
Xe 1955 0 93 53.2
Xe 1955 60 42 109
Au 2955 0 92 85.9
Au 2955 60 41 174
RHrFPGA Heavy Ion Testing at Room Temperature at
TAMU

• Orientation Test fixture was oriented so
  angular rotation was parallel to the gate width
  of the test devices transistors
• Angular rotation was limited to 60 degrees due to
  fixture shadowing of the die at higher angles of
  incidence

9
RHrFPGA Heavy Ion SEU Test Configuration at TAMU
10
RHrFPGA TEST PROGRAMS (Vectors)
Test Program Name Flip-Flops Tested Configuration Bits Tested
Application with I/O (Demodulator) 1506 131152
Load/Verify -Application 100 131152
Load/Verify Boot Zeros 100 131152
Full Shift Register Vertical Dynamic 1450 131152
Shift Register with Logic Dynamic 297 131152
Shift Register with Xbus Dynamic 184 131152
Shift Register with Lbus 670 131152
Shift Register w/Other Dynamic 975 131152
RHrFPGA Heavy Ion SEU Test Programs at TAMU
11
TEST PROCEDURE

• Establish the correct test conditions
• Load the RHrFPGA controller and test device with
  the proper configurations and verify test set
  functionality
• Irradiate the test device to the desired
  effective fluence while monitoring the device for
  SEU and monitoring for proper health
• Read the controller status registers to determine
  the number of upsets or test anomalies
• Read the test device configuration to check for
  configuration SRAM upsets
• Record all relevant test data from exposure run

12
HEAVY ION TESTING RESULTS

• The test evaluated the RHrFPGA using eight
  different test programs and configurations
• Seven were optimized for SEU testing to evaluate
  specific internal memory elements within the
  device and one test program represented a current
  RHrFPGA application
• Nominal supply voltage is 3.3V and the devices
  were tested at the worst case voltage of 3V
• Tests well below nominal (1.8V and 2.1V) were
  utilized to validate error detection
• The RHrFPGA test devices did not experience SEU
  or other SEE to the maximum available test LET of
  174 MeV/(mg/cm2)
• Performed at minimum rated supply voltage of 3.0V
• Applied all eight tests for fluences of ? 1.0x107
  ions/cm2 per test

13
PROTON TEST MOTIVATION AT IUCF

• Memory Elements The RHrFPGA memory elements use
  SEU hardening techniques similar, but harder than
  those of Honeywells HX6408 SRAM
• The HX6408 SRAM proton sensitivity was shown to
  be susceptible to the proton angle of incidence
  on the die
• Sensitivity attributed a single secondary heavy
  ion hitting two transistors within a memory cell
• The SEU cross-section is highest for a proton
  angle of incidence parallel to the path between
  the two sensitive transistors in a cell
• A grazing parallel proton beam oriented normal to
  this path is thus expected to produce an SEU
  cross-section similar to a normal incidence beam.
• The test irradiated the RHrFPGA with proton beam
  nearly parallel to the die surface
• The die was oriented so that the beam was
  parallel to the sensitive path direction

14
POSITIONING ASSEMBLY AT IUCF

• The rotation angle was limited to 70 degrees due
  to alignment constraints and concerns of
  irradiating the control device
• All exposures were performed at a 70 degree angle
  of incidence
• Different axes of rotation for the two cell types
  were used because the configuration RAM and
  application flip-flops are orthogonal to each
  other

Positioning Assembly at IUCF
15
RAM CELL TEST
Test Board in horizontal position with rotation
about the vertical axis
16
FLIP-FLOP TEST
Test Board in vertical position with rotation
about the vertical axis
17
RHrFPGA Proton SEU Test Configuration at IUCF
18
RHrFPGA TEST PROGRAMS (Vectors)
Test Program Name Flip-Flops Tested Configuration RAM Bits Tested
Application with I/O (Demodulator) 1506 131152
Full Shift Register Vertical 1450 131152
Shift Register with Lbus 670 131152
Shift Register with Logic 297 131152
Shift Register with Xbus 184 131152
5760 5790 131152
RHrFPGA Proton SEU Test Programs at IUCF
19
PROTON TESTING RESULTS

• The test evaluated the RHrFPGA using six
  different test programs and configurations
• The test was optimized to evaluate the RHrFPGAs
  two unique types of memory elements.
• The RHrFPGA test devices were irradiated to a
  proton fluence of 3.4x1013 p/cm2 at a beam energy
  of 203 MeV
• Test parts did not exhibit SEU or any other SEE,
  demonstrating that the RHrFPGA is essentially
  immune to proton-induced SEU

20
SUMMARY

• Heavy Ion Testing
• Two RHrFPGA devices did not upset to the maximum
  available test LET of 174 MeV/(mg/cm2) and a
  maximum ion fluence of ? 1.0x107 ions/cm2
• The test consisted of seventeen exposure runs at
  the minimum specified operating voltage of 3.0V
• The test results were consistent with analytical
  predictions indicating a much higher minimum SEU
  LET threshold than could be attained from a heavy
  ion SEU test
• Proton Testing
• Three RHrFPGA devices did not experience SEU or
  other SEE to a proton fluence of 3.4x1013 p/cm2
  per test device
• Test results were also consistent with analytical
  predictions indicating that the device is not
  sensitive to proton induced SEE

21
(No Transcript)
22
ACKNOWLEDGEMENTS
Kenneth A. LaBel Martha V. OBryan Gary Gardner
John Lintz
SPONSORS NASA Electronic Parts and Packing
(Program) Solar Dynamic Observatory (Project)