ASIC

Fault Models in VLSI

4 min read

Fault Models in VLSI

1. Simple Explanation (Gist)

Fault models in VLSI (Very Large Scale Integration) are simplified, abstract representations of physical defects that can occur in integrated circuits, used to predict how these defects will affect the circuit’s behavior and to develop effective testing strategies.

2. Detailed Breakdown

Why Fault Models are Needed

Physical defects in VLSI chips, such as shorts, opens, or manufacturing variations, are numerous and complex. It’s impractical to test for every possible physical defect directly. Fault models abstract these physical defects into logical representations, significantly reducing the number of faults to consider and making test generation and fault simulation feasible.

Common Fault Models

  • Stuck-at Fault Model: This is the most fundamental and widely used fault model. It assumes that a signal line or gate output is permanently “stuck” at a logic 0 (Stuck-at-0 or SA0) or a logic 1 (Stuck-at-1 or SA1), regardless of the circuit’s inputs.
    • Example: If a wire is stuck-at-0, it will always carry a logic ‘0’, even if the circuit tries to drive a ‘1’ onto it.
    • Significance: Its simplicity makes it easy to model and implement, allowing for efficient testing strategies.
  • Bridging Fault Model: This model represents a short circuit between two or more signal lines that should not be connected. These can occur due to manufacturing defects like metal debris.
    • Types:
      • Wired-AND Bridging: If two bridged nets have different values, the resulting value becomes a permanent logic 0 (like an AND operation).
      • Wired-OR Bridging: If two bridged nets have the same value, the resulting value remains unchanged (like an OR operation).
    • Significance: Bridging faults can create unexpected logic behavior and are often more challenging to detect than stuck-at faults.
  • Delay Fault Model: This model addresses defects that affect the timing performance of a circuit, causing signals to propagate slower or faster than intended. These faults may only manifest under specific operating conditions.
  • Transistor Stuck-On/Open Faults: These models focus on faults within individual transistors, the basic building blocks of VLSI circuits.
    • Stuck-On Fault: A transistor is permanently stuck in the conducting (on) state, regardless of its gate voltage.
    • Stuck-Open Fault: A transistor is permanently stuck in the non-conducting (off) state, acting like an open circuit.
    • Significance: These faults directly affect the transistor’s ability to switch, impacting the circuit’s logic behavior.

Fault Assumptions

  • Single Fault Assumption: Assumes that only one fault occurs in a circuit at a time. This simplifies testing but may not cover all real-world scenarios.
  • Multiple Fault Assumption: Considers that multiple faults might occur simultaneously in a circuit.

Fault Collapsing

Techniques like equivalence collapsing and dominance collapsing are used to reduce the total number of faults that need to be considered for testing, as some faults may produce the same faulty behavior or one fault’s detection implies another’s.

3. Conclusion

Fault models are essential tools in VLSI design and testing. By providing abstract representations of physical defects, they enable engineers to systematically analyze potential failures, generate efficient test patterns, and ensure the reliability and functionality of complex integrated circuits.

Further Reading

Graph View

Backlinks