SLLA541 December   2020 TCAN1144-Q1 , TCAN1146-Q1

 

  1.   Trademarks
  2. 1Introduction
  3. 2Hardware Component Failure Modes Effects and Diagnostics Analysis (FMEDA)
    1. 2.1 Random Fault Estimation
      1. 2.1.1 Fault Rate Estimation Theory for Packaging
      2. 2.1.2 Fault Estimation Theory for Silicon Permanent Faults
      3. 2.1.3 Fault Estimation Theory for Silicon Transient Faults
      4. 2.1.4 The Classification of Failure Categories and Calculation
    2. 2.2 Using the FMEDA Spreadsheet Tool
      1. 2.2.1 Mission Profile Tailoring Tab
        1. 2.2.1.1 Confidence Level
        2. 2.2.1.2 Geographical Location
        3. 2.2.1.3 Life Cycle
        4. 2.2.1.4 Use Case Thermal Management Control (Theta-Ja) and Use Case Power
        5. 2.2.1.5 Safe vs Non-Safe (Safe Fail Fraction) for Each Component Type
        6. 2.2.1.6 Analog FIT Distribution Method
        7. 2.2.1.7 Operational Profile
      2. 2.2.2 Pin Level Tailoring Tab
      3. 2.2.3 Function and Diag Tailoring Tab
      4. 2.2.4 Diagnostic Coverage Tab
      5. 2.2.5 Customer Defined Diagnostics Tab
      6. 2.2.6 Totals - ISO26262 Tab
      7. 2.2.7 Details - ISO26262 Tab
    3. 2.3 Example Calculation of Metrics
      1. 2.3.1 Assumptions of Use for Calculation of Safety Metrics
      2. 2.3.2 Summary of ISO 26262 Safety Metrics at Device Level

Fault Estimation Theory for Silicon Transient Faults

TI uses experimental data collected on process test chips to estimate silicon transient faults. Other data from vendors and foundries may also be used in this calculation, depending on the process technology used for the device. TI has been conducting targeted radiation exposure testing on process test chips since 2000 and is considered an industry leader in this area. TI's data correlates strongly to estimates for soft error provided in the International Technology Roadmap for Semiconductors (ITRS). At present, TI is not aware of any failure estimation standard that includes models to estimate FIT rate for transient faults.

Data taken on test chips has been utilized to establish base failure rates for single event upset (SEU) on SRAM bits and sequential digital logic. A further estimation is made for single event transient (SET) events for combinatorial logic. This failure mode is theoretically possible but TI has not been able to generate this failure mode in any testing done to date. ROM, analog, and package FIT have no contribution to transient faults and are therefore excluded from this calculation.

SEU failure rates consider exposure to two elementary particles: alphas and neutrons. Alpha particle exposure occurs primarily from radioactive material in the package mold compound. Low-alpha mold compound is utilized to minimize this failure rate. Neutron particle exposure is primarily due to cosmic particles bombarding the Earth. The altitude of operation and location on the Earth have impact on the rate of exposure, with high altitude locations near the equator having worst exposure. There is no effective way to manage neutron particles other than operation of the unit behind several feet of lead, water, or similar barrier. All of the estimations used in this report are based on JEDEC JESD89A, Measurement and Reporting of Alpha Particle and Terrestrial Cosmic Ray Induced Soft Errors in Semiconductor Devices, with assumption of neutron flux = 1 (measured exposure to neutrons as seen at sea level in New York City, USA).