A trace impedance target does not automatically define the impedance of a via transition. A via is a three-dimensional discontinuity involving a plated barrel, pads, antipads, reference planes, return vias, and any unused stub.
Why trace and via specifications differ
Trace impedance can often be evaluated as a transmission-line cross-section. A via transition must also account for the layer change, return-current path, and its behavior across the signal bandwidth. Define its performance with appropriate simulation and measurement rather than assuming a universal percentage tolerance.
1. Drilling, plating, and finished-hole dimensions
Drill-tool diameter and finished-hole diameter are different specifications. Plating and process allowances affect the finished opening. For example, a nominal 0.20 mm finished hole normally requires a larger drill-tool diameter, but the exact tool size and tolerance must be agreed with the fabricator.
Changing the barrel diameter can change transition capacitance, inductance, and loss. The amount depends on the full pad, antipad, stack-up, stub, and return-path geometry. Do not assign a fixed ohm change or tolerance percentage without the model, frequency range, and material data that produced it.
2. Comparing filled and unfilled vias
A useful comparison holds the conductive barrel, pad, antipad, stub, and reference-return geometry constant while varying the fill material. Include the fill dielectric properties and loss, or its electrical conductivity when a conductive fill is used. The conceptual cross-sections below illustrate the comparison; they do not represent measured or simulated results.
Evaluate the same return-loss and insertion-loss definitions, ports, bandwidth, mesh settings, and de-embedding for both models. Report the resulting differences with clear axes and units. Similar-looking curves alone do not establish that two structures are electrically identical.
What skin effect does and does not show
Skin effect concentrates high-frequency current near conductor surfaces, but it does not by itself prove that via-fill material has no influence on an entire transition. Field distribution and the surrounding geometry must be considered. A small effect in one design is not a universal result for all fill materials, frequencies, or via structures.