Eight microstrip traces of different lengths (0.35×–2.2× the base length) fan out from one driver pad to eight component pads, all sharing the same width, copper thickness and dielectric height. Each trace's electrical behaviour is computed from real geometry, not guessed:
R = ρ_cu · L / (W · t_cu) series resistance, ρ_cu = 1.68e-8 Ω·m
ε_eff = (εr+1)/2 + (εr-1)/2 · 1/√(1+12H/W) Hammerstad microstrip effective permittivity
v = c / √ε_eff propagation velocity along the trace
t_pd = L / v propagation delay (wavefront arrival)
C = ε0·ε_eff·W·L / H capacitance to the ground plane
τ = R·C RC time constant
V(t) = V_in·(1 − e^(−(t−t_pd)/τ)) voltage at the far pad after the wavefront arrives
The wavefront (bright dot) always reaches the far pad first, after t_pd — that part of the delay is fixed by the dielectric and the trace length, not by resistance. Only once the wavefront has arrived does the pad start charging toward the logic threshold along the RC exponential above; the glow you see ramping in is that charge, not the wavefront itself.
- Trace length / width / copper weight / dielectric height — each recomputes R, C, t_pd and τ for all eight traces from the formulas above.
- Clock rate — sets how often the driver re-fires. If a trace's total settle time (t_pd + 3τ, ≈95% charged) exceeds one clock period, that trace has not finished settling before the next edge arrives — its row in the table turns red ("Margin" goes negative) and its pad never reaches full brightness, a real high-speed-PCB timing-closure failure.
- Regime readout — a trace counts as a distributed transmission line (rather than a simple RC lump) when its length exceeds roughly v·t_rise/6; short, fat, well-spaced traces stay RC-dominated even at high clock rates.