main
John Lauer 0.10.4: 4-switch buck-boost converter model (issue 24); local impedance reference per section (issue 23) 781d6ef

adom-fields 0.10.4

  • 4-switch buck-boost converters (issue 24). The converter model no longer assumes a single-switch-node buck.
    • Through series parts: the inductor is found through low-value series parts on its path (a current-sense shunt, a ferrite, a 0 Ω link), or named in spec converter.inductor.
    • Switch legs: a switch leg is every part on its node (spec sources.<node>), so a discrete half-bridge's hot loop runs from the high side's input pin to the low side's ground pin.
    • Buck-boost: an inductor between two switch nodes is a 4-switch buck-boost.
      • Buck mode is evaluated at the input voltage: the input capacitors on the input leg.
      • Boost mode is evaluated at spec converter.vinMin: the output capacitors on the output leg, the inductor at Iout/(1−D).
      • Each mode reports its duty, inductor current and ripple, capacitor ripple, hot loop and emissions in review.converter.modes. Without vinMin, a note says what the boost mode needs.
    • Never null: a switch node that cannot be modelled now returns {"unsupported": "<why>"}, never null.
    • Unchanged elsewhere: a plain buck gives exactly the numbers and findings it gave before.
  • Impedance reference found locally (issue 23).
    • What changed: each trace segment's reference plane is now found under it. On each side, it is the nearest layer whose ground or supply copper covers 90 % of the segment (centre and both edges, every 0.25 mm). It used to be the nearest layer that a whole-layer statistic called a plane.
    • Every row names its reference: the layer, net, distance and coverage.
    • Partial references get no number: a nearer layer covering only part of a segment makes it "mixed reference", and nothing under it makes it "no reference"; neither gets an impedance.
    • Transitions stay separate: points within 0.3 mm of a via, a through-hole or the trace's own pads are left out of the coverage, and a segment that is all transition counts in transitionMm.
    • Target findings: they use only fully referenced sections, and a target net with unresolved sections gets its own finding.
    • The check: on a four-layer test board, a USB pair over a local In2 GND strip now reads the same 86.2 Ω whether or not unrelated In2 copper elsewhere crosses the whole-layer plane threshold. It read 110.9 or 86.2 before. With the strip broken, the pair is reported as mixed, not given an ideal-plane value.