Safety in LFSR/PRBS
Overview
Linear Feedback Shift Registers (LFSRs) and Pseudo‑Random Binary Sequence (PRBS) generators are widely used in digital systems for pattern generation, scrambling, link testing, and built‑in self‑test (BIST). In safety‑critical designs, they play a key role in LBIST, high‑speed interface validation, and data‑integrity mechanisms. Because LFSR/PRBS structures directly influence diagnostic coverage, their correctness and robustness are essential for functional‑safety compliance.
This page focuses on the safety implications, failure modes, and diagnostic strategies associated with LFSR/PRBS structures.
Where LFSR/PRBS Are Used
LFSR and PRBS generators appear in multiple safety‑relevant contexts:
Pattern Generation for LBIST
- Used as pseudo‑random stimulus for logic testing
- Sequence quality directly affects fault excitation and coverage
PRBS‑Based Link Testing
- SERDES and high‑speed interfaces rely on PRBS patterns for stress testing
- Detects bit errors, jitter sensitivity, and signal‑integrity issues
Scrambling and Whitening
- Used to decorrelate data streams
- Prevents long runs of identical bits
- Improves EMI and spectral distribution
Polynomial‑Based Mechanisms
- CRC
- Checksums
- Line coding
- Error‑detection and correction schemes
Built‑In Pattern Sources
- Protocol stress tests
- Interface bring‑up
- Manufacturing diagnostics
In all these cases, the quality and correctness of the pseudo‑random sequence directly affect diagnostic coverage and test effectiveness.
Safety‑Relevant Failure Modes
Wrong Polynomial
The implemented feedback polynomial does not match the intended one, reducing sequence length or diagnostic coverage.
Incorrect Seeding
Invalid, constant, or non‑diverse seeds limit randomness and reduce fault excitation.
Stuck or Illegal States
The LFSR enters an all‑zero or illegal state and stops toggling, producing a constant or short sequence.
Broken Feedback Path
Stuck‑at or bridged XOR gates or taps alter the sequence, reducing randomness and masking faults.
Determinism or Periodicity Issues
Unexpected repetition, shortened period, or correlation with system activity hides certain faults.
Generator/Checker Mismatch
PRBS generator and checker use different polynomials or seeds, causing:
- false error reports
- missed faults
- invalid link‑test results
These faults can silently degrade LBIST quality, invalidate link tests, or compromise safety assumptions in FMEDA.
Diagnostic Mechanisms
Signature Comparison in LBIST
- MISR signatures compared against golden references
- Detects incorrect LFSR behavior during startup or periodic tests
Online PRBS Monitoring
- PRBS checker compares received sequence with expected pattern
- Detects deviations, bit errors, or loss of randomness
Configuration Checks
- Verify polynomial, tap positions, and seed values at startup
- Prevents misconfiguration or unintended changes
Self‑Test Sequences
- Short, known sequences confirm that the LFSR can traverse expected states
- Detects lock‑up or illegal‑state behavior
Cross‑Checking
- Independent pattern sources or redundant checkers
- Detect failures in the primary LFSR/PRBS generator
Safety Architecture Considerations
Impact on LBIST Diagnostic Coverage
LFSR correctness directly affects the ability of LBIST to detect structural faults.
Credibility of PRBS‑Based Link Tests
Faulty PRBS generators or checkers invalidate link‑quality measurements.
FMEDA Assumptions
Sequence length, randomness, and diagnostic coverage must be justified in the safety case.
Verification Evidence
Safety‑critical designs require:
- polynomial validation
- seed‑diversity analysis
- fault‑injection results
- coverage reports
Sequence Quality
Maximal‑length polynomials and robust seeding strategies improve diagnostic confidence.
Related Technical Pages
DFT & BIST
Data Integrity
Coding & Randomness
- Safety in Line Coding
- LFSR / PRBS — Overview, Families & Architecture
- LFSR / PRBS — Mathematical Background
- LFSR / PRBS — Reference Table
