How does AOCV (path-depth/distance-dependent derating) reduce pessimism compared to flat OCV , with a worked numeric example?
From PDVerse STA Mentor Guide · pdVerse Mentor Guide
Technical Explanation
AOCV makes the derate factor depend on path characteristics — chiefly path depth (number of stages) and sometimes distance — instead of applying one flat derate everywhere, because statistically, variation partially cancels out over many stages rather than accumulating at the same rate as a single-stage flat multiplier would suggest.
Timing-path Reasoning
A flat OCV derate applies the same percentage multiplier regardless of how many stages the path traverses, which overstates variation on long, deep paths where random perstage variation partially averages out. AOCV instead uses a derate table indexed by path depth, applying a smaller percentage the deeper the path, recovering pessimism that flat OCV would otherwise bake in.
Equation
20-stage path, 1000ps nominal: flat OCV -> 1080ps (+80); AOCV (depth
20, +2.8%) -> 1028ps (+28); 52 ps of pessimism recovered. Lookup
table: depth 1/3/6/10/15/20 -> derate 8.0/6.5/5.2/4.1/3.3/2.8%. Path
A (3 stages, derate 6.5%): 1000 x 1.065 = 1065ps. Path B (15 stages,
derate 3.3%): 1000 x 1.033 = 1033ps.Expected Result
Shorter/shallower paths get a higher derate percentage (since fewer stages give less statistical cancellation), while longer/deeper paths get a progressively lower derate percentage, more accurately matching physical variation behavior.
Failure Symptoms
Flat OCV signoff on deep paths reports larger, artificially pessimistic delay adders (e.g., +80ps vs AOCV's +28ps on the 20-stage example) that can produce false violations not reflective of actual silicon risk.
Root Cause
Flat OCV treats derate as a constant percentage irrespective of path depth, ignoring that per-stage random variation does not simply linearly stack across many independent stages.
Mentor Note — Debugging Procedure
Confirm the AOCV table is indexed correctly by depth (and distance where applicable) and cross-check reported delay adders against the expected derate percentage for that path's actual stage count. AOCV's depth-indexed derate table converts a blunt flat-percentage assumption into a path-aware one, recovering tens of picoseconds of pessimism on deep paths — as the 20-stage example shows, 52ps of margin that flat OCV was needlessly consuming.
Continue learning free
Get a practical low-power chapter
Receive the existing “Low Power and Multivoltage Fundamentals” PDF chapter and its download link by email.
Continue practising