Measuring Clock Fanout Skew at the Required Reference Points

A clock buffer’s output-skew specification describes particular device measurement points and conditions. The skew seen by two destination devices also includes routing, loads and measurement error. Define the reference points before comparing a board-level result with a component data sheet.

Choose the timing quantity

The LMK1C110x data sheet distinguishes output-to-output skew, part-to-part skew, propagation delay and additive jitter. These are different quantities; a small value for one does not establish the others.

State whether the requirement concerns two outputs on one buffer, outputs on different buffers or the clock arrival times at destination pins. Specify the active edge and threshold used to define time. Include the load and supply conditions that apply to the comparison.

Build a path-delay budget

List the buffer contribution, PCB and cable delays, termination effects and destination loading for each path. Compare differences between paths rather than simply adding every absolute propagation delay. Retain separate allowance for uncertainty where component guarantees or physical dimensions are incomplete.

For illustration, two traces with identical drawn lengths can still have different delay if they use different structures or loads. Conversely, a small difference at the buffer package may be outweighed by an intentional routing delay. The acceptance criterion should apply at the points the receiving system actually uses.

Remove measurement asymmetry

Use matched probes or cables and characterize the timing difference between acquisition channels. Swap channels or probes as a cross-check when practical. Record the deskew procedure and avoid treating the instrument’s displayed time difference as inherently free of channel delay.

Check how edge amplitude and slew rate affect threshold crossing. A different probe load on one output can both change the circuit and shift the measured crossing. Keep the observation arrangement as symmetric as the required accuracy demands.

Separate mean offset from variation

Capture enough edges to characterize the relative timing distribution under the intended operating conditions. Report the mean or static offset separately from cycle-to-cycle variation using clearly defined metrics. Do not relabel an RMS jitter result as a worst-case static skew guarantee.

Retain the reference-point drawing, load conditions, instrument correction and representative distributions. Include relevant temperature and supply conditions in the verification plan. When routing or destination parts change, revisit the board contribution even if the clock buffer remains identical. This produces a timing statement that is meaningful to the receiving devices and prevents comparison of two numbers measured at different places under incompatible definitions. Record the threshold convention beside each reported timing result.