The clock reaches each flip-flop through real wires and buffers, so the same edge arrives at different flip-flops at slightly different times. That difference is clock skew.
This course uses one sign convention:
tskew = (when the edge reaches the capturing flip-flop) − (when it reaches the launching flip-flop)
Positive skew means the capturing flip-flop's clock is late. Some books use the opposite sign, so check which one a question uses.
Skew shifts both timing checks, in opposite directions:
- Setup: a late capturing edge gives slow data more time.
T ≥ tcq + tpd + tsu − tskew. - Hold: a late capturing edge also keeps its hold window open longer, so fast data is more likely to break in.
tcq(min) + tcd ≥ th + tskew.
So positive skew helps setup and hurts hold; negative skew does the reverse. Chip designers build balanced clock trees to keep skew small, and a gated clock is a classic way to create skew by accident.