How does crosstalk change delay, not just add noise?

Short Answer

When the aggressor and victim switch at the same time, the coupling capacitance changes how much charge the victim's driver must move. If they switch in opposite directions, the victim sees roughly its own capacitance plus twice the coupling, so it slows down. If they switch the same way, it speeds up, so crosstalk can hurt setup on one path and hold on another.

Technical Explanation

• A coupling capacitor between two wires has one wire on each plate. What the victim driver has to charge depends on how much the voltage across that capacitor changes.
• If only the victim switches, the coupling capacitor sees one full swing and acts like capacitance to ground: Cv + Cx.
• If the aggressor switches the opposite way at the same time, the voltage across the capacitor changes by two swings, so the effective load is about Cv + 2Cx. The victim transition slows and its delay grows.
• If both switch the same way, the capacitor sees almost no change and the victim moves faster than normal.
• The slowdown matters for setup on max paths, and the speed-up matters for hold on min paths. Signal integrity analysis applies each where it hurts.
• Timing windows decide whether the aggressor can actually switch while the victim does. If their windows never overlap, the delta is filtered out, which avoids false pessimism.
• Fixes include more spacing, shielding, changing layers, upsizing the victim driver, or making aggressor and victim drive strengths more similar.

Formula Or Decision Rule

• Opposite-direction switching: the victim sees an effective load of about Cv + 2Cx, so it slows down (positive delta delay).
• Same-direction switching: the coupling capacitor sees little voltage change, so the victim speeds up (negative delta delay).
• Cv = victim capacitance to ground, Cx = coupling capacitance to the aggressor.

What To Check

• The crosstalk delta column in timing reports for the worst setup and hold paths.
• Nets with the largest delta delay, and who their aggressors are.
• Whether the aggressor and victim timing windows really overlap.
• Whether signal integrity analysis was enabled when the timing was run.

Healthy, Suspicious & Hard-stop Results

• Healthy (illustrative): Delta delay on critical paths of a few picoseconds, with slack covering it in every scenario.
• Suspicious (illustrative): One net carrying most of a path's delta delay, usually a long parallel run beside a busy bus.
• Hard stop: Timing signed off with signal integrity analysis turned off.

What The Interviewer Is Testing

• Whether you can explain the Cv + 2Cx result physically.
• Do you apply crosstalk to both setup and hold?
• A strong answer shows you know timing windows filter out impossible cases.

Follow-up Question & Model Response

"Why does a stronger victim driver reduce delta delay?"

Candidate Model Response: Delta delay comes from extra charge the victim driver has to move during its transition. A stronger driver has lower output resistance, so it moves that extra charge faster, and the transition stretches less. It also holds the victim more firmly against the aggressor's push. That is why upsizing the victim driver is a standard fix, although it adds input load to the previous stage.