What is diagnostic coverage?

Diagnostic coverage is one of the most important numbers in functional safety engineering — and one of the least understood outside the field.

IEC 61508 defines diagnostic coverage (DC) as the fraction of the total rate of dangerous hardware failures that are detected by automatic diagnostic tests. It's expressed as a percentage.

A safety system with 60% diagnostic coverage detects 6 out of every 10 dangerous hardware faults through automatic self-testing. A system with 99% DC detects 99 out of 100. That gap — between 60% and 99% — is the engineering difference between a system that can support SIL 1 and one that can support SIL 3.

IEC 61508-2:2010, Table A.2 defines three ranges. Low diagnostic coverage spans 60% to less than 90%. This level is sufficient for SIL 1 architectures. Medium covers 90% to less than 99%, required for SIL 2. High means 99% or greater, required for SIL 3.

Why this matters for Physical AI: diagnostic coverage is a critical input to determining whether a safety architecture can achieve a given safety rating. In IEC 61508, DC feeds into the Safe Failure Fraction (SFF), which — combined with Hardware Fault Tolerance (HFT) — determines the SIL the architecture can claim. In ISO 13849, average diagnostic coverage (DCavg) is used directly as one of the factors determining the Performance Level. Either way, insufficient DC limits your achievable safety rating regardless of how many redundant channels you build.

Achieving high diagnostic coverage requires specific techniques: comparison of redundant channel outputs, monitoring of watchdog timers, memory tests (walking bit, checksum), I/O comparison, cross-monitoring between processors, and power supply monitoring. Each technique contributes a quantifiable improvement to the overall DC calculation.

The question functional safety asks isn't "does your safety system work?" It's "what percentage of its failures can your system detect before they cause harm?" Diagnostic coverage is how that question gets answered.